We publish only what can be checked against the original discussion.
Current Issue · Published
AI Coding Tools
Issue 01 · Published September 2026
CLAIM C001
Supported
2026-09-30
Public discussion records debate whether adopting tool-specific configuration like CLAUDE.md rather than AGENTS.md causes developer ecosystem fragmentation.
Full record
CLAIM C002
Supported
2026-10-02
GitHub issue #57580 in anthropics/claude-code, opened 2026-05-09 by JDL747, records a report that long Bash-heavy Claude Code Desktop sessions accumulate PTY master file descriptors against /dev/ptmx until macOS's kern.tty.ptmx_max (511) is exhausted, after which forkpty(3)/posix_openpt(3) fail with ENXIO for every process on the machine; the reporter notes that quitting and relaunching Claude releases the descriptors.
Full record
CLAIM C003
Supported
2026-10-03
GitHub issue #97854 in anthropics/claude-code, opened 2026-09-28 by timothe-rey, records a report that during a Claude Code session with Auto mode active, the server-side safety classifier intermittently returned no verdict for consecutive tool calls including Bash and ScheduleWakeup, resulting in total failure before self-resolving; subsequent in-thread discussion through October 2026 records further user reports across macOS, Windows, Linux, and WSL debating whether the failure is transient or requires terminating the session.
Full record
CLAIM C004
Supported
2026-10-04
GitHub issue #46917 in anthropics/claude-code, opened 2026-04-12 by Adrian-Mteam, records a report that Claude Code versions 2.1.100 and 2.1.101 consume approximately 20,000 more cache_creation_input_tokens per request than v2.1.98 despite sending fewer bytes in the client request payload, characterizing the difference as server-side token inflation; subsequent thread discussion records user verification and discussion of workarounds.
Full record
CLAIM C005
Supported
2026-10-05
GitHub issue #88747 in anthropics/claude-code, opened 2026-08-22 by DearMaestro, records a report that worktree creation writes an absolute core.hooksPath into config.worktree, causing hooks fired from a worktree to execute scripts from the main checkout on whatever branch it happens to be on; the report characterizes a weakened or removed hook step still passing as a silently removed safety net.
Full record
CLAIM C006
Supported
2026-10-06
GitHub issue #25128 in anthropics/claude-code, opened 2026-02-12 by emregurhan, records a report that drag and drop is fully operational in terminal/CLI mode but completely non-functional in the VS Code extension chat panel, a regression introduced around v2.1.6 and still unresolved as of v2.1.39; the issue consolidates multiple fragmented reports into a single tracking issue.
Full record
CLAIM C007
Supported
2026-10-07
GitHub issue #50270 in anthropics/claude-code, opened 2026-04-18 by tpdhd, records a report that starting with v2.1.113 Claude Code switched from a JavaScript entry point to a native glibc Linux binary, completely breaking it on Termux (Android); users see a "native binary not installed" error and are locked out of updates.
Full record
CLAIM C008
Supported
2026-10-10
GitHub issue #68619 in anthropics/claude-code, opened 2026-06-15 by lonlundgren, records a report that subagents recursively spawn child agents 50+ levels deep while ignoring the CLAUDE_CODE_FORK_SUBAGENT=0 setting, burning 1.2M+ tokens in ~30 minutes on a task that should have been a git clone; the report characterizes the behavior as a major regression shipped on or around June 10th.
Full record
CLAIM C009
Supported
2026-10-17
GitHub issue #22543 in anthropics/claude-code, opened 2026-02-02 by mjbyt, records a report that after using the cowork feature Claude Desktop becomes extremely slow, alongside a VM bundle file that grows to 10GB, is never cleaned up, and regenerates quickly after deletion; the report documents a cleanup test showing approximately 75% faster performance immediately after removing the bundle, while also noting performance degrades again over time even with the bundle at 0 bytes, suggesting a memory leak or accumulating work.
Full record
CLAIM C010
Supported
2026-10-12
GitHub issue #5140 in anthropics/claude-code, opened 2025-08-05 by siraj-samsudeen, records a report that user-level permissions in ~/.claude/settings.json are not enforced during command execution even though they appear as loaded in /permissions; the same permissions work when placed in project-level .settings.local.json.
Full record
CLAIM C011
Supported
2026-10-13
GitHub issue #11315 in anthropics/claude-code, opened 2025-11-09 by nenrightld-ux, records a report of a severe memory leak in which Claude Code consumed 129GB of virtual memory, exhausted all 16GB of system RAM, and caused a complete system freeze requiring a hard reboot; the report documents physical memory growing from ~2GB to ~12GB in 30 minutes with no other resource-intensive applications running.
Full record
CLAIM C012
Supported
2026-10-16
GitHub issue #85891 in anthropics/claude-code, opened 2026-08-11 by kylealty-boop, records a report that on Windows 11 the Claude Desktop main window behaves as a topmost window, staying drawn above other applications even after another window is focused, with no in-app setting to turn this off; the reporter ruled out third-party pinning tools and concludes the topmost flag appears to originate from the app itself.
Full record
CLAIM C013
Supported
2026-10-17
GitHub issue #22264 in anthropics/claude-code, opened 2026-01-31 by ilanoh, records a report that when Claude Code makes multiple tool calls in parallel in a single message and one of them fails, all remaining sibling calls in that batch are automatically cancelled with the error "Sibling tool call errored", forcing a retry of all the cancelled calls individually.
Full record
CLAIM C014
Supported
2026-10-19
GitHub issue #1455 in anthropics/claude-code, opened 2025-05-31 by jennifgcrl, records a report that Claude Code writes its cache data and configs to ~/.claude.json and ~/.claude instead of following the XDG Base Directory specification; the reporter notes CLAUDE_CONFIG_DIR can redirect some files but states the default behavior without configuration should follow the XDG specification.
Full record
CLAIM C015
Supported
2026-10-20
GitHub issue #87640 in anthropics/claude-code, opened 2026-08-18 by sagimedina1, records a report that the Fable 5 `[reasoning_extraction]` safeguard rejected a message whose user text was the single word `Hi`; the reporter states the classifier guards against prompts attempting to extract chain-of-thought, that nothing in the conversation did that, and that the same session context works normally on Opus 5, pointing at Fable-only safeguards rather than at the conversation.
Full record
CLAIM C016
Supported
2026-10-22
GitHub issue #2805 in anthropics/claude-code, opened 2025-07-01 by pm0code, records a report that on Ubuntu Linux, Claude Code creates shell scripts and text files with Windows line endings (CRLF) instead of Unix line endings (LF) even when CLAUDE.md carries explicit instructions; the reporter documents "No such file or directory" errors when executing the scripts and notes a manual sed conversion workaround.
Full record
CLAIM C017
Supported
2026-10-23
GitHub issue #74260 in anthropics/claude-code, opened 2026-07-04 by federbenjamin, records a report that on Claude Code 2.1.201 with adaptive thinking, assistant text emitted mid-turn — a text block followed by another thinking block or further tool calls in the same turn — is frequently neither rendered on any surface nor persisted to the session transcript JSONL; in one measured session, 12 of 13 mid-turn status texts vanished from both the display and the JSONL.
Full record
CLAIM C018
Supported
2026-10-23
GitHub issue #8856 in anthropics/claude-code, opened 2025-10-03 by Sundeepg98, records a report that Claude Code creates temporary files at /tmp/claude-{random-4-hex}-cwd to track working directory changes across Bash command executions but never deletes them, causing accumulation of /tmp/claude-*-cwd files; the reporter observed 174 files accumulated in one day of usage.
Full record
CLAIM C019
Supported
2026-10-23
GitHub issue #7134 in anthropics/claude-code, opened 2025-09-04 by edlyra, records a report that Claude Code does not respect the original file encoding when editing files, assuming UTF-8 for all operations, which causes character corruption when working with files encoded in other formats such as Windows-1252; the reporter documents using the Edit tool on text with Portuguese accented characters (configuração, não) in a Windows-1252 file.
Full record
CLAIM C020
Supported
2026-10-28
GitHub issue #75043 in anthropics/claude-code, opened 2026-07-07 by mof086999-code, records a report, based on a controlled experiment of 10+ orchestrator runs over two days, of three related failures when an agent spawned via the Agent tool spawns its own children via the Agent tool: children run detached regardless of run_in_background, child completion notifications reach the main conversation rather than the spawning subagent, and TaskStop fails with ownership errors after a SendMessage resume.
Full record
CLAIM C021
Supported
2026-10-28
GitHub issue #93782 in anthropics/claude-code, opened 2026-09-12 by kryan007, records a report that after auto-updating to 2.1.269, text inserted by a voice dictation tool (Wispr Flow, clipboard plus simulated Ctrl+V) is no longer inserted into the Claude Code prompt when Claude Code runs in the VS Code integrated terminal (Remote-WSL), while 2.1.268 in the exact same terminal works.
Full record
CLAIM C022
Supported
2026-10-28
GitHub issue #97074 in anthropics/claude-code, opened 2026-09-25 by qvarker, records a report, based on a controlled A/B test, that on a Max subscription the same workload uses about 1.8 times more of the 5-hour window when it runs headless via claude -p (entrypoint sdk-cli) than in the interactive TUI (entrypoint cli), with token usage the same in both modes.
Full record
CLAIM C023
Supported
2026-10-28
GitHub issue #8327 in anthropics/claude-code, opened 2025-09-29 by Toowiredd, records a report that users with valid Claude Pro/Max subscriptions receive a "This organization has been disabled" error in the Claude Code CLI, with the reporter attributing it to a set ANTHROPIC_API_KEY environment variable (from a previously disabled organization) overriding the valid subscription authentication.
Full record
CLAIM C024
Supported
2026-10-29
GitHub issue #64479 in anthropics/claude-code, opened 2026-06-01 by wodin, records a report that the bug from #52813 is still reproducible: the Edit tool fails when a multi-line old_string spans a region where the same Unicode character appears as both a literal codepoint and a source escape, and the whole-string swap fallback tries all literals and all escapes but cannot handle the mixed case.
Full record
CLAIM C025
Supported
2026-10-29
Hacker News comment item 50279944, posted 2026-10-28 by InsideOutSanta in the thread for the story The Agent SDK, records that Anthropic employees clarified in HN comment threads that the API credits are in addition to the existing behavior and not a replacement, and that other harnesses can still call the Claude Code CLI and use the subscription quota that way.
Full record
CLAIM C026
Supported
2026-10-29
GitHub issue #52871 in anthropics/claude-code, opened 2026-04-24 by danielvos1998, records a report that the Claude Code MCP OAuth client appends a trailing slash to the resource parameter on both the authorize redirect and the token exchange, so that when the metadata of an MCP server declares the resource URI without a trailing slash, Microsoft Entra ID rejects the request with AADSTS9010010.
Full record
CLAIM C027
Supported
2026-10-29
GitHub issue #78857 in anthropics/claude-code, opened 2026-07-18 by ankitdotgg, records a report that after updating to Claude Code Desktop 1.22209.0, the sidebar filter menu in the Code tab no longer offers a project filter, leaving no way to limit the sidebar to a chosen set of projects, while the official docs still describe filtering by project.
Full record
CLAIM C028
Supported
2026-10-30
GitHub issue #80988 in anthropics/claude-code, opened 2026-07-24 by elaye-canopy, records a report that Claude Code 2.1.219 injects a system-prompt section registered internally as heron_brook whose default text tells the model not to call the AgentTool and not to use workflows or deep-research unless the user requested it, that the section is enabled by model capability for Opus 5 rather than by user configuration, and that it offers no documented opt-out, with the reporter saying it overrode an explicit standing delegation instruction in CLAUDE.md.
Full record
CLAIM C029
Supported
2026-10-30
Hacker News comment item 50288423, posted 2026-10-29 by Asyakguen in the thread for the story Claude Code regularly sends each background agent’s transcript to its model, records a comment that asks whether almost all tokens would be cache hits if the entire transcript were sent for a summary, and says that the 16.4 percent figure comes from the Claude Code cost_usd estimate, so cheap per token still adds up when it is 1 in 5 requests.
Full record
CLAIM C030
Supported
2026-10-31
GitHub issue #102132 in anthropics/claude-code, opened 2026-10-14 by luca-derosa, records a report that in Desktop Code tab sessions a wrap_up call with an ask draws a next-step box that ends the turn, with nothing checking that the reply was written before the call, so in one evening session three wrap_up calls with an ask came before the answer and the user saw only a two-sentence recap and two buttons, and that a second wrap_up call in the same turn silently replaced the first box.
Full record
CLAIM C031
Supported
2026-10-31
GitHub issue #14202 in anthropics/claude-code, opened 2025-12-16 by ianaswanson, records a report that a plugin installed with project scope in one project shows as installed in the Marketplaces view of other projects and cannot be installed there, while the Installed tab in the other project correctly does not list it, with the reporter attributing the mismatch to the Marketplaces indicator and the install command not checking projectPath.
Full record
CLAIM C032
Supported
2026-10-31
GitHub issue #86014 in anthropics/claude-code, opened 2026-08-12 by lilysttail, records a report that cross-session send_message returned a success result to the sender while the message was never delivered to the target session and stayed loading in the desktop UI, with 4 of 4 sends in one evening showing success to the sender and 0 of 4 ever received by the target, in both busy and idle target states.
Full record
CLAIM C033
Supported
2026-10-31
GitHub issue #42765 in anthropics/claude-code, opened 2026-04-02 by alzerid, records a report that the Claude Code OAuth client uses a localhost callback URI on an ephemeral port when authenticating with MCP servers, which the reporter says violates RFC 8252 Section 7.3 because native apps must use loopback IP literals, so authorization servers that strictly follow the RFC reject the redirect URI for lack of dynamic port matching.
Full record
CLAIM C034
Supported
2026-10-31
GitHub issue #61172 in anthropics/claude-code, opened 2026-05-21 by eliasjamesbermudez, records a report that running /clear in the CLI terminal clears conversation history but does not reset the session name, so the post-clear session is created with the same name as the old one and both sessions appear under the same name in /resume with no way to distinguish them, reported as a regression from version 2.1.89 to version 2.1.146.
Full record
CLAIM C035
Supported
2026-10-31
GitHub issue #79366 in anthropics/claude-code, opened 2026-07-20 by sandopolus, records a report that starting a new session with worktree isolation enabled placed the session inside an existing worktree directory created by a previous session for an unrelated task, with a new branch checked out into the old directory, so the directory name no longer matched the task and leftover files from the previous task could leak into the new task context or pull request.
Full record
CLAIM C036
Supported
2026-11-02
GitHub issue #92984 in anthropics/claude-code, opened 2026-09-09 by tomokuri8, records a report that after the September 2026 Windows 11 cumulative security update KB5124008 was installed and the machine rebooted, the Cowork local sandbox in Claude Desktop for Windows could no longer mount any host drive share, with every 9p mount inside the guest failing with invalid argument, and that uninstalling KB5124008 and rebooting fully restored the sandbox.
Full record
CLAIM C037
Supported
2026-11-02
GitHub issue #27242 in anthropics/claude-code, opened 2026-02-20 by scapeshift-ojones, records a report that Claude Code has no working way to review conversation history that has been cleared or compacted, that the full data is always preserved in transcript.jsonl but the TUI provides zero functional paths to access it, covering post-compaction history, post-plan-mode clear context, and branch navigation, with manually parsing raw JSONL files as the only workaround.
Full record
CLAIM C038
Supported
2026-11-02
Hacker News comment item 50325704, posted 2026-11-01 by airhangerf15 in the thread for the story Ask HN: What is a good open-weight alternative to Claude Code, records a comment that corrects the terminology, saying Claude Code has no weights and is a harness that contacts Anthropic servers and runs their closed weight models, that the harness provides a means for the LLM to write file output and run tools to test output, and that OpenCode can be pointed at an Anthropic account only with API usage.
Full record
CLAIM C039
Supported
2026-11-05
GitHub issue #29937 in anthropics/claude-code, opened 2026-03-02 by efloehr, records a report that during normal Claude Code sessions in tmux the terminal rendering becomes corrupted, with text from different parts of the session overlapping and overwriting each other on the same lines so the interface becomes unreadable, that the corruption persists until the session is restarted, and that the only reliable fix is /clear or restarting the session.
Full record
CLAIM C040
Supported
2026-11-06
GitHub issue #62072 in anthropics/claude-code, opened 2026-05-24 by Nubaeon, records a report that after auto-update to Claude Desktop 1.8555.2, Cowork sessions on macOS have zero MCP tools bound across all four install paths, that the Cowork VM spawner produces zero --mcp-config or --allowedTools args across 500 or more spawn events, that no .mcp.json is written into the session dirs, and that the Chat tab is unaffected with the same MCPs initializing cleanly there.
Full record
CLAIM C041
Supported
2026-11-06
GitHub issue #96326 in anthropics/claude-code, opened 2026-09-23 by noboru007, records a report that in a long agentic session the model sent several user-facing replies in English despite a CLAUDE.md rule and a session language setting requiring Japanese, that the English replies always followed turns whose tool output was English, and that three corrections did not hold, with the model slipping back into English within the same turn or the turn after.
Full record
CLAIM C042
Supported
2026-11-07
GitHub issue #57084 in openai/codex, opened 2026-10-26 by interconnectedMe, records a report that on Linux / VS Code with ChatGPT Pro authentication, recurring "Selected model is at capacity" and "exceeded retry limit, last status: 429 Too Many Requests" errors interrupted ordinary Codex work on October 25-26 despite 84-99 percent weekly allowance remaining, that several October 25 retries converged on common deadlines within 0.7 seconds, and that the precise rejecting condition remains unknown.
Full record
CLAIM C043
Supported
2026-11-07
GitHub issue #58190 in zed-industries/zed, opened 2026-05-31 by martin-c, records a report that on a 64 GB M1 Max Mac Studio running Zed 1.4.4 stable on macOS 15.6.1, a single Zed process grew to 391 GB resident and drove the whole system into terminal memory pressure, that the growth coincided with a continuous main-thread hang lasting at least about 2 hours, and that the reporter argues no single subsystem should be able to consume several times physical RAM and take down the OS.
Full record
CLAIM C044
Supported
2026-11-07
GitHub issue #50428 in openai/codex, opened 2026-10-02 by mattalberta, records a report that Windows Codex Desktop can read an existing durable/cloud chat bound to a Windows execution environment, but direct plain-text submissions and a same-directory fork fail before execution with "invalid turn/start params: AbsolutePathBuf deserialized without a base path", that the recorded JSON-RPC code was -32602 with failureReason invalid_config, and that the same failure appeared in desktop versions 26.928.31416 and 26.930.21537.
Full record
CLAIM C045
Supported
2026-11-07
GitHub issue #67606 in anthropics/claude-code, opened 2026-06-11 by gisstw, records a report that two independent Opus 4.8 sessions on the same day exhibited severe confabulation in long-context sessions of about 100-170k tokens, that in one session the model acknowledged a user message that does not exist anywhere in the session transcript and then claimed the session was under a sustained prompt-injection attack, and that in the other session it asserted a production log line count that appears nowhere in any tool output and inverted which machine was production.
Full record
CLAIM C046
Supported
2026-11-07
GitHub issue #85856 in anthropics/claude-code, opened 2026-08-11 by momom2, records a report that on Windows with Git Bash every run of backslashes in a Bash tool command is halved before bash parses anything, that quoting cannot prevent it, and that the reporter attributes it to a mismatch between the MSVCRT quoting rules Claude Code uses to encode its command line and the different convention the MSYS2 bash binary uses to decode it, with Claude Code 2.1.222 on Windows 11 Home and Git for Windows bash 5.3.15.
Full record
CLAIM C047
Supported
2026-11-07
GitHub issue #50168 in openai/codex, opened 2026-10-02 by IaHorrorLabs, records a report that Dot can list and read Codex Cloud tasks but the exposed cloud_threads write path fails for both task creation and follow-up messaging, that new cloud task creation returned an app-server backend failure with error_code UNKNOWN and no usable thread or turn identity, and that a follow-up message to a manually created healthy cloud thread returned CloudThreadNotFoundError with error_code NOT_FOUND, observed on October 1, 2026.
Full record
CLAIM C048
Supported
2026-11-07
GitHub issue #13437 in cline/cline, opened 2026-08-20 by rvnlord, records a report that a Cline session fails on every run_commands call with an Invalid input error because the model omitted the documented command key and emitted the object-command shape under a guessed key, and that even when the command runs, the chat preview still shows the object placeholder instead of the command text, with Cline VS Code Extension 4.1.9 (Next), VS Code 1.132.0, PowerShell 7, and cline-pass with deepseek-v4-flash.
Full record
CLAIM C049
Supported
2026-11-07
GitHub issue #30408 in openai/codex, opened 2026-06-28 by kkkayye, records a report that the Codex app-server spawns a full set of global MCP server processes for each new thread or conversation but never kills them when threads are archived or closed, and that after 25 threads the reporter observed 133 orphaned MCP processes consuming about 9.3 GB RSS, with Codex Desktop 0.142.3 on macOS 15.x with 64 GB RAM.
Full record
CLAIM C050
Supported
2026-11-07
GitHub issue #62716 in zed-industries/zed, opened 2026-08-16 by ahjota, records a report that starting a Factory Droid thread from the Agent Panel in Zed v1.15.0 on macOS 14.7.2 (arm64) exits with status 1, that the package bin resolves to a 120MB Mach-O native binary after a postinstall step replaces the JS shim with a hardlink to the native platform binary, and that running the bin directly works while launching it through Node fails, with the expected behavior that Zed execute the resolved bin path directly.
Full record
CLAIM C051
Supported
2026-11-07
GitHub issue #51524 in openai/codex, opened 2026-10-07 by wogkek2540-ux, records a report that on October 6, 2026 previously working Windows Computer Use stopped being able to inspect browser windows in Chrome and Edge, that app and window enumeration succeeds but a window-state call stops the turn before any click or typing with a URL-confidence policy message, and that reopening windows, restarting the app, and fully rebooting the computer did not resolve it.
Full record
CLAIM C052
Supported
2026-11-07
GitHub issue #50915 in openai/codex, opened 2026-10-04 by Kizron, records a report that on Windows 11 build 26200 with Codex Desktop 26.930.3930.0 and bundled codex-cli 0.160.0, the MXC sandbox fails before launching any child process when an unrelated BitLocker data volume is present but locked, that the workspace is on another drive and does not reference the locked volume, and that MXC exits with a locked-volume enumeration error before the command starts.
Full record
CLAIM C053
Supported
2026-11-07
GitHub issue #35259 in openai/codex, opened 2026-07-24 by dimasyankauskas, records a report that during intentional Ultra and multi-agent work on Codex Desktop 26.721.31836 with codex-cli 0.146.0-alpha.3.1 on macOS, the desktop repeatedly re-entered the model merely to wait for agents or poll terminal status, that in a corrected reset-to-49-percent usage window wait and status polling turns accounted for 19.8 percent of raw local token volume, and that the polling often ran at a 30- or 60-second cadence for hours while delegated work was still running.
Full record
CLAIM C054
Supported
2026-11-07
GitHub issue #50769 in openai/codex, opened 2026-10-04 by jabdori, records a report that during October 3-4, 2026, development coordinated through dots hit repeated pre-execution approval blocks although the user had authorized development and publication, that later denials kept citing read-only scope or missing authorization, that a branch-creation call returned a user-cancelled message the user said they had not triggered, and that moving the workflow to gh CLI on the connected Mac still left branch creation rejected before execution, on macOS with GitHub CLI 2.101.0 and unknown Codex, client and backend builds.
Full record
CLAIM C055
Supported
2026-11-07
GitHub issue #103376 in anthropics/claude-code, opened 2026-10-19 by EthanRCT, records a report that on Claude Max (20x), the weekly all-models meter went from 0% to 54% in the first 22 hours after the weekly reset, while the reporter's Claude Code token volume in those 22 hours was nearly identical to the same 22 hours of the previous week, when about 6 times that volume moved the meter only to 60%, measured from local transcripts on Claude Code 2.1.299 on macOS 26.6.2.
Full record
CLAIM C056
Supported
2026-11-07
GitHub issue #95364 in anthropics/claude-code, opened 2026-09-18 by semikolon, records a report that when an update is staged, Claude Desktop 2.2553.1 on macOS 26.6.2 waits for a short idle window and then quits and relaunches itself, logging the restart as a stealth update, that the relaunch stops every running Code session and drops each session's Remote Control link, and that the behaviour appears to have started around 27 August 2026, with nine restarts in 22 days in the retained logs.
Full record
CLAIM C057
Supported
2026-11-08
GitHub issue #38455 in openai/codex, opened 2026-08-13 by flannick, records a report that ChatGPT desktop 26.810.41047, build 6570, on macOS 15.7.7, Apple Silicon, 32 GB RAM, repeatedly spawns Computer Use workers and crashes with a V8 out-of-memory error, that the previous version 26.730.61639 worked, that the crash reproduces while idle 98 seconds after launch through SIGABRT in node::OOMErrorHandler, with 316 threads at crash, 187 named computer-use, and telemetry showing 78 unexplained child processes consuming about 4.96 GB after 29 seconds, and that the Browser and Computer Use flags were explicitly disabled.
Full record
CLAIM C058
Supported
2026-11-08
GitHub issue #13365 in cline/cline, opened 2026-08-19 by gaogage, records a report that clicking Generate Commit Message in the SCM panel fails with [Commit Generation Failed] stdout maxBuffer length exceeded whenever the diff output exceeds the Node default exec buffer of 1MB, that the diff calls in apps/vscode/src/utils/git.ts go through the promisified exec and execFile wrappers without a maxBuffer option, verified on current main and in the shipped v4.1.10 bundle, and that the later truncation to 500 lines and 5000 chars happens only after exec has already killed the child process, so collecting the full stdout just to discard most of it is what breaks the feature.
Full record
CLAIM C059
Supported
2026-11-08
GitHub issue #34260 in openai/codex, opened 2026-07-20 by RocStone, records a report that Codex Desktop 26.715.4045.0 on Windows 25H2 x64, build 26200.8875, can enter an unbounded process-cleanup loop where hundreds of taskkill.exe processes remain alive, each with a corresponding conhost.exe, that the accumulated taskkill instances repeatedly query Win32_Process, exhaust the WMI provider quota, and make the entire Windows UI sluggish or unable to launch new processes, with peak measurements of 457 taskkill.exe processes, 473 conhost.exe processes, 1,388 total processes, over 14,000 threads, over 350,000 handles, and WMI provider private memory of 720.5 MB, and that killing the accumulated taskkill.exe processes immediately reduced the total process count from 1,388 to about 565 while WMI private memory fell to about 104 MB, but taskkill.exe counts grew again from 156 to 320 in 20 seconds while the original Codex process remained running.
Full record
CLAIM C060
Supported
2026-11-08
GitHub issue #15417 in cline/cline, opened 2026-10-28 by JoaoMarcos44, records a report verified against main commit 51f63e4fd63aeeb74d708b9c3d1f8b800aff9b3e on Windows, Node 22.16.0 and Vitest 4.0.18, that two editor calls replacing different regions of one file can both report success while one change disappears from disk, with a deterministic regression in which both calls return Edited but the final file can contain alpha new while beta stays old, and that replaceInFile and insertInFile in sdk/packages/core/src/extensions/tools/executors/editor.ts perform separate asynchronous read and full-file write operations with no coordination between editor executor instances.
Full record
CLAIM C061
Supported
2026-11-08
GitHub issue #15574 in cline/cline, opened 2026-11-06 by mwielgosz-esky, records a report on the VSCode Extension version 4.1.26, with behaviour identical on current main, that per-server autoApprove lists for MCP tools are accepted by the settings schema and read by the hub layer, but no code path consumes them when an MCP tool call is approved or prompted, so there is no configuration that makes a listed MCP tool run without prompting while another MCP tool prompts; the verified chain includes McpHub.ts reading each server autoApprove array and stamping a per-tool flag, a complete per-tool toggle disabled behind SHOW_MCP_PER_TOOL_AUTO_APPROVE set to false, buildToolPolicies forcing every discovered MCP tool to autoApprove false, and isToolAutoApproved resolving every MCP tool through the global useMcp setting.
Full record
CLAIM C062
Supported
2026-11-08
GitHub issue #92210 in anthropics/claude-code, opened 2026-09-04 by PedroGiudice, records a report that in Claude Desktop 1.46388.3 on Windows 11, opening a claude://code/new?folder= deep link for a folder equal to the folder already selected in the composer starts the session in a freshly minted scratch workspace instead of that folder, that the same link works when the selected folder is different or when the path is spelled differently, and that the reporter reproduced the failure 4 times in a row. The issue was closed 2026-11-06 by maintainer bpasero-ant, who stated it was fixed in Claude Desktop 4.16336.0; no independent fix verification appears in the thread.
Full record
CLAIM C063
Supported
2026-11-08
GitHub issue #45163 in openai/codex, opened 2026-09-13 by lightninglu10, records a report that in codex-cli 0.154.0 on macOS, Darwin 25.6.0, arm64, after the terminal changes between light and dark appearance while Codex is already running, the composer and previously rendered user-message blocks retain their old background colors, that the terminal default foreground updates, producing nearly white text on a pale background, so long-lived sessions become unreadable after an automatic macOS appearance change, that refocusing and redrawing do not refresh the cached palette, and that restarting refreshes the startup colors but the next appearance change can break the session again.
Full record
CLAIM C064
Supported
2026-11-08
GitHub issue #84863 in anthropics/claude-code, opened 2026-08-07 by bvdeenen, records a report that in Claude Code 2.1.224 on Linux, the sandbox.filesystem config exposes only an allowWrite allowlist with no read-restriction primitive, so credential directories such as ~/.ssh and ~/.aws are fully readable through Bash even with the sandbox enabled, that the agent added an undocumented sandbox.credentials key and later left settings.json invalid JSON with a trailing comma, after which the CLI kept running with no crash, warning, or blocked tool use, and that the agent then read ~/.ssh/id_rsa in a verification command while the sandbox was reported as enabled.
Full record
CLAIM C066
Supported
2026-11-08
GitHub issue #103788 in anthropics/claude-code, opened 2026-10-20 by kornai, records a report that on macOS 27.0, Claude Desktop 2.42304.2 or 3.1623.0 fails sign-in with a browser-verification error on both the Continue with Google path (400 POST /api/auth/verify_google) and the Continue with email path (400 POST /api/auth/send_magic_link), that the failure repeats after removing cookies, moving the whole profile aside, clean reinstall, and excluding the app from the VPN, that sign-in to claude.ai in Safari on the same machine works, and that switching the VPN to a relay in a different country let the email path through while the Google path still returned 400.
Full record
CLAIM C067
Supported
2026-11-08
GitHub issue #13380 in cline/cline, opened 2026-08-19 by D-Alex, records a report on Cline CLI 3.0.55 that an interactive session stops producing output mid-task and never returns control to the prompt, with no error and no timeout, that hub-daemon.log shows for 4 sessions running the identical prompt a capability.request.start for hook.afterRun with no matching capability.respond, staying open until the client process is killed, and that the identical prompt hung 4 out of 4 times while a control run with the missing file added to the repo completed normally.
Full record
CLAIM C068
Supported
2026-11-08
GitHub issue #15561 in cline/cline, opened 2026-11-05 by Lucifergene, records a report comparing Desktop App 0.0.49 with bundled cline-cli 3.0.74 against VSCode Extension 4.1.26, that when the Anthropic provider is pointed at a custom base URL, the Desktop App and CLI request the base URL plus /messages while the VSCode Extension requests the base URL plus /v1/messages, that both surfaces read from the same ~/.cline/data directory so one stored value cannot satisfy both, and that a local HTTP stub repro observed POST /messages where /v1/messages was expected, confirmed against a real Anthropic-compatible gateway where POST to /v1/messages returned 200 and POST to /messages returned a gateway error page.
Full record
CLAIM C069
Supported
2026-11-08
GitHub issue #13384 in cline/cline, opened 2026-08-19 by rvnlord, records a report on VS Code 1.132.0 with Cline extension saoudrizwan.claude-dev-4.1.9 (Next) on Windows 10/11, that search_codebase returns files from directories the user has already excluded in their ignore files such as generated build output, that in the ripgrep fast path the custom ignore files .clineignore and .agentignore are never read by ripgrep at all and .gitignore is only honored in git-based contexts, and that in the JS fallback walker there is no ignore-file parsing anywhere in the tool path.
Full record
CLAIM C070
Supported
2026-11-08
GitHub issue #35420 in openai/codex, opened 2026-07-26 by hiroki-tamba-research, records a report that in the ChatGPT Work / Codex surface on Windows, requests repeatedly fail with stream disconnected before completion, that the Windows workspace being used is OneDrive-backed and the OneDrive client is in a degraded state because it reports that the Microsoft 365 subscription payment could not be processed, that Codex cannot complete work in the selected workspace, and that the evidence establishes a strong correlation between a OneDrive-backed workspace, a degraded OneDrive state, and repeated stream disconnections, while not yet proving which layer is the root cause.
Full record
CLAIM C071
Supported
2026-11-08
GitHub issue #82074 in anthropics/claude-code, opened 2026-07-28 by uningeniero, records a report that the official Claude for Chrome extension can remain authenticated after Log out of all devices while its OAuth grant is not shown in the Claude Code token list or the normal active-session inventory, that a separate user independently reproduced the sequence in which the web session, Claude Code, and Windows and Android apps were logged out while Claude for Chrome remained authenticated and usable, and that Anthropic confirmed through HackerOne that Claude for Chrome uses a separate OAuth authorization and that the revocation gap was already known internally before this report. The report itself states the evidence does not yet prove a stolen token caused the unauthorized usage in the authors account, and attribution remains a hypothesis. The issue was closed 2026-09-07 by github-actions[bot] for inactivity, not as a fix; no fix confirmation appears in the thread.
Full record
CLAIM C072
Supported
2026-11-09
GitHub issue #14853 in cline/cline, opened 2026-10-06 by dominiccooney, records a report that an audit of
the docs against the shipped products recorded 201 findings, that the issue tracks fixing them in 18 PRs,
one per product area, and that of the 201 findings, 138 are ready to write, 59 need a quick look in the
product while the PR is written, and 4 need no edit.
Full record
CLAIM C073
Supported
2026-11-09
GitHub issue #14759 in cline/cline, opened 2026-10-02 by dominiccooney, records a report that twenty
images and videos in the docs show UI that has since changed, that thirteen no longer match the product
while the other seven have outdated labels around content that is still right, and that each item in the
issue says what is wrong and where the new file goes, so anyone with the right setup can take one. The
issue remains open.
Full record
CLAIM C074
Supported
2026-11-09
GitHub issue #11370 in cline/cline, opened 2026-06-08 by galaticpower, records a report that the Cline
v0.0.43 hub daemon generates a random 32-char hex auth token on startup regardless of whether the client
has any auth configured, that when no CLINE_HUB_AUTH_TOKEN env var is set the client connects with no auth
so the server random token does not match and the connection is rejected with code 1006, and that in the
default no-auth configuration the hub should accept connections where both server and client have an empty
token. The issue was closed 2026-11-08 by linear-code[bot]; no fix confirmation appears in the thread.
Full record
CLAIM C075
Supported
2026-11-09
GitHub issue #11369 in cline/cline, opened 2026-06-08 by galaticpower, records a report that Cline v0.0.43
introduced a regression in the hub daemon where WebSocket connections are rejected when the authToken is
an empty string, that the XB auth-token comparison function rejects connections when the sec-websocket-protocol header contains an empty string for the auth token prefix, and that WebSocket connections should
be accepted when no auth token is configured. The issue was closed 2026-11-08 by linear-code[bot]; no fix
confirmation appears in the thread.
Full record
CLAIM C076
Supported
2026-11-09
GitHub issue #89690 in anthropics/claude-code, opened 2026-08-26 by urda, records a report that a
modelPicker row with model opusplan never reaches the /model picker in append mode because Claude Code
treats the row as covered by the built-in lineup, while the built-in lineup carries no Opus Plan Mode row
in an ordinary session, and that replace mode renders the opusplan row, which the author states proves the
row parses, validates, and is servable. The author retested the report on later builds through 2.1.319 and
reported the same result in append mode. No maintainer response appears in the thread.
Full record
CLAIM C077
Supported
2026-11-09
GitHub issue #59401 in zed-industries/zed, opened 2026-06-16 by chrrity, records a report that large tool
outputs from grep and read_file are sent without truncation in Zed agent threads, that accumulated raw
tool outputs exceeded 1M tokens by the fourth or fifth user message in the reported thread, that the
thread JSON was 13.3 MB uncompressed with 99.7 percent raw tool execution output, and that a single grep
call returned 5.2 million characters which gets stored and re-sent on every subsequent request. yeskunall
replied in the thread that a lot of what was identified had actually been addressed since version 1.6.3,
and the author clarified in a comment the same day that the source code analysis was on a stale local
checkout, while stating the crash itself is real and the remaining gap is grep and read_file lacking
output truncation.
Full record
CLAIM C078
Supported
2026-11-09
GitHub issue #15579 in cline/cline, opened 2026-11-06 by The-time-forever, records a report that a saved
Cline CLI 3.0.74 session cannot be resumed from /history because the Hub connection closes with code 1006
and reason Connection ended, that the Hub log records repeated abnormal socket closures with reason
Received too big message, and that a session file of about 23.4 MiB containing 775 messages is a likely
match for the affected session. The author states the socket-closure log entries carry no session ID, so
the association cannot be confirmed from the logs alone.
Full record
CLAIM C080
Supported
2026-11-09
GitHub issue #100730 in anthropics/claude-code, opened 2026-10-09 by TheAviv, records a report that in an
attended Claude Code cloud session the auto mode classifier refused a fire_trigger call on the account
owner's own scheduled task, with the reason Cowork Run Routine Now, and refused or stopped a file transfer
between two machines the owner controls as Data Exfiltration, and that there is no setting, prompt or
override to approve the refused action. The author added further instances in comments through 2026-11-08.
No maintainer response appears in the thread.
Full record
CLAIM C081
Supported
2026-11-09
GitHub issue #15594 in cline/cline, opened 2026-11-08 by rjvkn, records a report that the Cline CLI splits
a single assistant message into separate transcript rows when the provider interleaves reasoning chunks
with text chunks, so a Thinking row is injected between the two halves of one sentence and the answer
reads as if it was cut off, while the model output itself is intact. The report traces the split to use-agent-events.ts starting a new transcript entry every time the stream switches type, and notes the VS Code
task path already handles interleaved reasoning while the CLI handler has no equivalent gate.
Full record
CLAIM C082
Supported
2026-11-09
GitHub issue #54685 in openai/codex, opened 2026-10-16 by n0rpan123, records a report that a naturally
required cross-task execution review denied a bounded QA installation because the trusted user context
contained only an abbreviated Approved with omitted context, even though the exact approval request
contained the complete original proposal and reply, original message identifiers and hashes, and a read-only original-record locator. The final decision was risk level high, user authorization unknown, outcome
deny, and the report asks maintainers whether this is a trusted-context retention or attribution defect, a
reviewer evidence-verification failure, or a currently unsupported cross-task authorization path.
Full record
CLAIM C084
Supported
2026-11-09
GitHub issue #54434 in anthropics/claude-code, opened 2026-04-28 by KingPin, records a report that in
long-running interactive Claude Code sessions SSE responses from /v1/messages repeatedly stop emitting
events mid-response without sending message_stop or any error event, that most stalls recover on their own
after about 50 to 70 seconds while some never recover and only resolve when the user presses ESC, and that
5 distinct stalls occurred in a single 26-minute working session. The issue was closed 2026-09-08 by
github-actions[bot] for inactivity, not as a fix, and a commenter the same day stated it was incorrectly
closed; no fix confirmation appears in the thread.
Full record
CLAIM C085
Supported
2026-11-09
GitHub issue #63751 in anthropics/claude-code, opened 2026-05-29 by Call-me-Boris-The-Razor, records a
report that Claude Code blocked a request to design anti-tamper and code-hardening for the author's own
proprietary desktop application with a Usage Policy and cyber-safeguard error, and that after the first
block, unrelated benign follow-up messages in the same session were also blocked, including messages with
no technical content at all. The report states the classifier appears to score the accumulated
conversation context rather than the individual message, so a single false positive poisons the whole
session, and that the Cyber Verification Program appeal path is not a working remedy for independent
developers.
Full record
CLAIM C086
Supported
2026-11-10
GitHub issue #15427 in cline/cline, opened 2026-10-29 by peperonid, records a report that the
standalone cline-core service (JetBrains plugin) crashes repeatedly with a JavaScript heap out of
memory error (exit code 134) after roughly 6-7 minutes of uptime, that heap snapshots show
hundreds of retained duplicate copies of the fully serialized globalState JSON, that the on-disk
globalState.json is only 2.5 KB so the 11.6 MB in-memory serialization is itself abnormally large,
and that the same crash pattern repeated 3 times in 25 minutes; a later investigation note in the
issue states the same root cause as issue #15031, the JetBrains host-bridge state stream queuing
every snapshot without backpressure and never dropping a superseded one.
Full record
CLAIM C087
Supported
2026-11-10
GitHub issue #41220 in openai/codex, opened 2026-08-27 by FromAriel, records a meta-issue tracking
report that many open Codex reports describe a common symptom family in which Codex subscription
quota or purchased credits are depleted materially faster than the user's prior baseline or local
and token evidence would predict, sometimes with abrupt changes during an unchanged usage window,
passive or background consumption, or usage records that do not reconcile with the meter, and that
the tracker is intended to link and compare those reports in one place while not assuming they all
have one root cause.
Full record
CLAIM C088
Supported
2026-11-10
GitHub issue #74558 in anthropics/claude-code, opened 2026-07-05 by randalmurphal, records a
report that on Fable 5, when the model emits thinking, text and tool_use within a turn, the client
intermittently never receives the text block and instead receives an extra short thinking block
with a valid signature whose content is a paraphrase of the missing user-facing text, so affected
turns look completely silent, and that 5 of 7 text-bearing tool-calling turns in the session above
were affected.
Full record
CLAIM C089
Supported
2026-11-10
GitHub issue #15031 in cline/cline, opened 2026-10-12 by mibangjie38-bit, records a report that
cline-core running inside Android Studio grows until it hits the 8 GB heap limit and dies, killing
the session mid-task, that the heap is 98 percent strings and it is one payload copied 1111 times,
that a single suspended generator frame holds all 1111 payloads, and that the message translator
cannot map session_snapshot to anything and logs and drops it while the producer keeps buffering.
Full record
CLAIM C092
Supported
2026-11-10
GitHub issue #14059 in cline/cline, opened 2026-09-11 by Energytechglobal, records a report that
when multiple Cline terminals run in the same workspace, a shared relay state file can associate
requests with the wrong task, that the OpenAI-compatible request does not receive a request-scoped
x-litellm-trace-id header from Cline, and that when a shared relay or state workaround is used,
the most recently captured task ID can overwrite the other terminal's ID, causing requests to be
attributed to the wrong session; the issue proposes adding the header from the current
request-scoped session ID.
Full record
CLAIM C093
Supported
2026-11-10
GitHub issue #13346 in cline/cline, opened 2026-08-18 by rvnlord, records a report that
run_commands (Background Console) output reaches the LLM verbatim, ANSI escape codes included,
that run_commands spawns the command with piped stdio and forwards stdout and stderr unprocessed
into the tool result, and that Cline bundles strip-ansi-compatible code only inside table-layout
dependencies and never applies it to command output; current behaviour is that output contains
literal ANSI sequences.
Full record
CLAIM C095
Supported
2026-11-10
Hacker News comment item 50430601, posted 2026-11-09 by ariwilson in the thread for the story Is
Opus 5.5 nerfed? Thinking patterns changed on November 3, records a re-run of the author's own
baseline evaluation on Opus 5.5 at medium effort in Claude Code, that 20 of the tasks were last
run on Nov 1, twice each, that Nov 1 solved 36 of 40 and today solved 19 of 20, and that the one
significant change was speed, with median time to first token decreasing from 3.0s to 1.7s; the
comment notes the caveat of one run per task today, medium effort only.
Full record
CLAIM C096
Supported
2026-11-10
Hacker News story item 50422628, posted 2026-11-09 by DenGorbachev, records a report that ChatGPT
Desktop forwards the user's SSH agent to remote machines, that the agent running on a remote
machine can authenticate using the user's local SSH credentials, and that it can push to the
user's private repos on GitHub, for example; the story cites as proof a discussion in openai/codex
pull request 28131.
Full record
CLAIM C099
Supported
2026-11-11
GitHub issue #65051 in anthropics/claude-code, opened 2026-06-03 by vertliba, records a report that in background (daemon) sessions on Claude Code 2.1.161 (Linux), assistant text blocks are not persisted to the session transcript whenever the response also contains tool_use blocks and two thinking blocks (interleaved thinking), that the transcript ends up with only thinking records plus the tool_use records with the text record missing entirely, that this is a regression from 2.1.160 where every text block was persisted correctly, and that after the restart on 2.1.161 there were zero assistant text records other than the first response across about 10 consecutive responses that all contained user-facing text, while text-only final responses persist fine.
Full record
CLAIM C100
Supported
2026-11-11
GitHub issue #25178 in openai/codex, opened 2026-05-30 by Define1165250535, records a report that on Windows Codex Desktop on Windows 10 Pro 22H2 (build 10.0.19045), any get_window_state call that requests a screenshot fails before capture with "SetIsBorderRequired failed: 不支持此接口 (0x80004002)", that the same target window works when requesting accessibility text only, and that the reporter traces the failure to the capture helper calling GraphicsCaptureSession.IsBorderRequired / SetIsBorderRequired on a Windows build that does not support that property, which Microsoft documents as introduced in Windows 10 version 2104 / build 10.0.20348.0.
Full record
CLAIM C101
Supported
2026-11-11
GitHub issue #59955 in zed-industries/zed, opened 2026-06-26 by lixiaoyan, records a report that Zed always displays stale cached content for files under .git instead of the version on disk, demonstrated by writing "a" to .git/COMMIT_EDITMSG, opening it in Zed, closing the tab, writing "b" to the same file, and reopening it in Zed; that the issue significantly affects users who configured VISUAL="zed --wait", since every commit or interactive rebase can potentially show stale file content; and that Zed should always display files under .git using the latest content. The report was filed against Zed v1.9.0+preview on macOS 26.5.1.
Full record
CLAIM C102
Supported
2026-11-12
GitHub issue #58447 in openai/codex, opened 2026-10-30 by Robert-Fish, records a report that a dot cannot read an existing user-created native Codex thread on a connected macOS desktop through the documented cloud_threads.read route, despite the desktop reporting connected, attached, and authorized for tasks; that a control read of an existing dot-created cloud task succeeds; and that the read fails with DesktopThreadReadUnavailableError (CONFLICT / desktop_thread_read_unavailable), with no root cause asserted.
Full record
CLAIM C103
Supported
2026-11-12
GitHub issue #55200 in zed-industries/zed, opened 2026-04-29 by jakeleventhal, records a report that after uninstalling an external agent (OpenCode / opencode), Zed still opens a new agent thread with opencode selected in the agent picker, though the picker only shows Zed Agent and Codex CLI as available options, and the thread fails immediately with "Failed to Launch / Custom agent server `opencode` is not registered"; the report was filed against Zed 0.233.10 on macOS 26.5.
Full record
CLAIM C104
Supported
2026-11-12
GitHub issue #59855 in openai/codex, opened 2026-11-06 by ShakhrukhDR, records a report that opening an affected cloud-managed chat from the Codex Desktop sidebar on Windows can show "Failed to resume chat / Invalid request: AbsolutePathBuf deserialized without a base path" before a new message is submitted, while reading the stored conversation can still succeed; the report asks Desktop to resolve workspace paths when resuming cloud-managed chats from Windows and notes which field triggers the error remains unknown.
Full record
CLAIM C105
Supported
2026-11-12
GitHub issue #14045 in cline/cline, opened 2026-09-10 by lamian521, records a report that in Cline CLI 3.0.61-nightly, fresh session/new sessions work normally, but after restoring a session with two completed turns in a new process, session/load succeeds and replays the history, yet the immediately following session/prompt emits only session_info_update and resolves with end_turn, zero new agent_message_chunk events and zero body characters, so the requested token is not recalled; the report rules out the post-load model pin as the cause, since both variants failed identically.
Full record
CLAIM C106
Supported
2026-11-12
GitHub issue #15641 in cline/cline, opened 2026-11-10 by Frederic-Demaison, records a design report that when using Cline against a local llama.cpp inference server with prompt/KV caching enabled, every Plan-to-Act toggle forces a large partial or full prompt re-prefill, measured at 17-36 s and 24k-46k tokens, even though the conversation history is unchanged, because the system prompt embeds a mode-conditional instruction block and the tool set differs between modes, with the editor hidden in Plan mode; the report proposes making the serialized prefix byte-stable across mode switches and supplies a working patch for v4.1.26 with before/after measurements.
Full record
CLAIM C107
Supported
2026-11-12
GitHub issue #53015 in zed-industries/zed, opened 2026-04-02 by m-ar-c, records a report that exporting a 468 MB SQL file at the root of a project caused Zed to use a huge amount of RAM and crash the reporter's GNOME session, reproduced as soon as Zed opened because the Uncommitted Changes tab was restored, and reproducible just by opening that tab, with RAM continuing to grow after closing it, and the problem gone if the file is removed; the report was filed against Zed v0.230.0 on Linux Wayland Debian sid with 15.3 GiB of memory.
Full record
CLAIM C108
Supported
2026-11-13
GitHub issue #42695 in openai/codex, opened 2026-09-04 by HS7097, records a report that a task delivered through `send_message_to_thread` on Codex Desktop on Windows (bundled CLI 0.153.0) was received and partially executed correctly, but after automatic mid-turn compaction the receiving agent reverted to an older, already-completed instruction instead of finishing the current task; that its final response then incorrectly reported that no task execution had occurred, despite successful tool results and files created earlier in the same turn; that the incoming Task B instruction was recorded as a `function_call_output` associated with `send_message_to_thread` while the replacement history retained the older Task A instruction as a visible user message; and that the opaque compaction item's contents cannot be inspected, so the report does not establish that Task B was entirely absent from the compressed summary. The report notes whether this shares a root cause with related reports requires investigation, and no root cause is asserted.
Full record
CLAIM C109
Supported
2026-11-13
GitHub issue #39054 in openai/codex, opened 2026-08-17 by rmanalan, records a report that when an MCP server rejects a refresh with a spec-correct `400 invalid_grant`, Codex does not discard the refresh token, re-sends the identical token on every subsequent attempt indefinitely, and never falls back to a fresh authorization flow; that the app layer is simultaneously told the server is still authenticated, so no re-authenticate affordance appears; that the user sees the MCP server present but with zero tools; that the report was reproduced on Codex CLI 0.140.0, 0.144.0, 0.146.0, 0.147.0 and 0.148.0-alpha.20 on macOS 26.5.2 arm64; and that the report attributes the behavior to the credential being judged usable purely because a non-empty refresh-token string is present, with the rejected refresh token never recorded. No root cause beyond the reporter's stated attribution is asserted.
Full record
CLAIM C110
Supported
2026-11-13
GitHub issue #25770 in openai/codex, opened 2026-06-02 by zyzhen, records a report that Codex Desktop from Microsoft Store repeatedly fails to update on Windows 10 Enterprise (version 10.0.19044, x64); that the update reaches the AppX/MSIX registration phase and then fails because Windows still considers the old `OpenAI.Codex` package/application to be in use, with AppXDeployment-Server failures including `0x80073CF9`, `0x8007001F` and `0x80073D02`; that this happened repeatedly on the same machine, and also on another machine in the same LAN; that closing visible Codex windows is not enough, since Windows still reports the old package/app as in use; and that the update succeeded only after manually quiescing the old package state and reinstalling from Microsoft Store. The report notes the failure appears to be in the Windows Store/MSIX update lifecycle around releasing the old package identity, rather than in normal application runtime behavior after the package is successfully installed. No root cause beyond the reporter's stated attribution is asserted.
Full record
CLAIM C111
Supported
2026-11-13
GitHub issue #14069 in cline/cline, opened 2026-09-11 by JimRomaineHRI, records a report that in the Cline JetBrains plugin 2.0.4, on a light IntelliJ theme, the plan-mode message banner header text and the copy icon render yellow text on a yellow background and are effectively invisible on light themes; that the report attributes this to the `--warning` custom property being static and not included in the per-theme variables IntelliJ injects onto the body, unlike `--background`, `--foreground` and `--input-*`, causing `text-warning/70` on `bg-warning/10` to fail contrast on light IntelliJ themes; that the VS Code extension does not appear to exhibit this issue; and that the static `--warning` fallback value works on dark themes but fails on light ones. No root cause beyond the reporter's stated attribution is asserted.
Full record
CLAIM C112
Supported
2026-11-13
GitHub issue #11678 in anthropics/claude-code, opened 2025-11-15 by ecarx-ai-adm, records a report that the `claude -p` command (headless mode) fails immediately with API Error 400 when attempting to execute any prompt, that the error reports `tools.3.custom.input_examples: Extra inputs are not permitted`, that interactive mode works fine and only headless mode is affected, and that the report was filed against Claude Code 2.0.42 on macOS (Darwin 25.1.0) with model claude-sonnet-4-5-20250929, with no root cause asserted.
Full record
CLAIM C113
Supported
2026-11-13
GitHub issue #50520 in openai/codex, opened 2026-10-03 by gwdonnor, records a report that the native Windows sandbox repeatedly fails before shell commands start with `helper_sandbox_lock_failed` when opening the `.codex\.sandbox-bin` directory, that granting SYSTEM and Administrators Full Control on that directory and retrying Windows setup restores sandbox access but the problem later recurs with the ACL entries again at Modify, that this is an observed recurrence with a successful workaround and not a confirmed diagnosis of which component resets the ACL, and that the report was observed October 1-2, 2026 with command-runner 0.160.0 in workspace-write filesystem mode, with no root cause asserted.
Full record
CLAIM C114
Supported
2026-11-13
GitHub issue #8043 in zed-industries/zed, opened 2024-02-20 by kvark, records a report that most ordinary pixels are being overdrawn 5-6 times on screen, that the green area shows 5 draws per pixel and the yellow area on the left 6 draws per pixel, that this burns GPU power unnecessarily and drains the battery, and that ideally these areas would have 0 to 1 draw, with the report filed from Linux.
Full record
CLAIM C115
Supported
2026-11-13
GitHub issue #39523 in anthropics/claude-code, opened 2026-03-26 by interconnectedMe, records a meta-issue report that `bypassPermissions` mode does not bypass permissions, that the problem has been broken in various forms since July 2025 (#2933) and remains broken on v2.1.84 (March 2026), that v2.1.78 introduced hardcoded protection for `.claude/`, `.git/`, `.vscode/`, and `.idea/` directories that overrides every bypass mechanism, and that the permission mode spontaneously downgrades from Bypass permissions to Edit automatically during a session, with the report mapping two correlated failure branches and a duplicate trail across linked issues, and no single root cause asserted.
Full record
CLAIM C116
Supported
2026-11-13
GitHub issue #93482 in anthropics/claude-code, opened 2026-09-10 by GBalunis, records a report that in a cloud Cowork session linked to a desktop device (Claude Desktop 1.49585.0, Windows 11 25H2, claude-opus-5), when device_commit_files writes back to a file that already exists on the linked device, it returns written with no rejections and the file's modification time advances to the moment of the call, but the content on disk is the payload from the previous commit; committing the identical payload a second time lands it; creating a new file is unaffected and correct every time; the report calls this a silent data-integrity failure, rules out a read cache, OneDrive, KB5124008, and the force flag, and offers a suggested area to look at without asserting a root cause.
Full record
CLAIM C117
Supported
2026-11-13
GitHub issue #15663 in cline/cline, opened 2026-11-11 by carlosabella, records a report that in Cline 4.1.26 (VSCode Extension, macOS Sequoia 15.7.4), the skills tool description sent to the model lists only skill names, while the description from each SKILL.md frontmatter is parsed but never shown to the model until after it has already invoked the skill; as a result the model can only pick a skill by guessing from its name, and anything with a less obvious name is effectively only reachable by typing the skill name manually; the report locates the behavior in createSkillsTool in sdk/packages/core/src/extensions/tools/definitions.ts, which maps configuredSkills to names only although SkillsExecutorSkillMetadata already carries description, and says the reporter has a small fix ready plus an updated unit test.
Full record
CLAIM C118
Supported
2026-11-13
GitHub issue #57101 in openai/codex, opened 2026-10-26 by americanmademischief, records a report that on Windows with Codex desktop package OpenAI.Codex 26.1021.3705.0 and Computer Use plugin 26.1021.31345, a gold/yellow border remained around the display after a Codex chat had finished its native desktop capture work, with the Windows Computer Use helper still running and its Windows capture-access record still open approximately five hours after the screenshot-producing turn completed; the report asks for capture resources to be released when their owning work ends while preserving Computer Use for other chats that actually need it, and labels the lifecycle-ordering explanation a hypothesis, not a confirmed root cause.
Full record
CLAIM C119
Supported
2026-11-14
GitHub issue #42501 in openai/codex, opened 2026-09-03 by leonfetter, records a report that after updating the Microsoft Store Codex Windows app to 26.901.1978.0, Codex launched several ChatGPT.exe processes but never created a visible window, with all processes responsive but MainWindowHandle remaining 0; investigation showed the new bundled cua_node runtime repeatedly failed during staging, each launch creating a new incomplete staging directory that contained node.exe but was consistently missing node_repl.exe, and manually staging the packaged cua_node runtime with xcopy /G into the expected versioned runtime directory fixed the problem immediately, after which Codex launched normally; the report presents the EFS / Application Protected WindowsApps relocation issue as the apparent cause, not a confirmed root cause.
Full record
CLAIM C120
Supported
2026-11-14
GitHub issue #15690 in cline/cline, opened 2026-11-12 by Jingyi-A-Chen, records a report that in Cline CLI 3.0.74 (also seen on 3.0.73), if the model calls spawn_agent with a task that is not a string, the whole CLI process exits with code 1 with TypeError: input?.trim is not a function; the report traces the cause to spawn_agent declaring task as z.string() without validating the input before execute runs, the session callback starting handleSubAgentStart through an unawaited void invokeBackendOptional call so its rejection escapes the try/catch, and deriveTitleFromPrompt calling input.trim() on the object, with the unhandled rejection handler then ending the process; the expected behavior stated in the report is that a wrong-typed argument is answered with a tool error and the session continues.
Full record
CLAIM C121
Supported
2026-11-14
GitHub issue #29908 in openai/codex, opened 2026-06-24 by legka, records a report that on Ubuntu 24.04 with Bubblewrap 0.9.0, apply_patch and ordinary managed sandbox commands fail before the requested command can run, with the error bwrap: loopback: Failed RTM_NEWADDR: Operation not permitted; the report attributes the failure to Bubblewrap sandbox setup, not repository file permissions, noting minimal Bubblewrap tests fail with uid-map and loopback errors under AppArmor unprivileged user namespace restrictions, and asks for the apply_patch and managed sandbox path to be made robust on Ubuntu systems with such restrictions or for a supported fallback that does not require changing host-wide security settings.
Full record
CLAIM C122
Supported
2026-11-14
GitHub issue #28058 in openai/codex, opened 2026-06-13 by ignatremizov, records a report that after #26210 (Encrypt multi-agent v2 message payloads, merged 2026-06-05, post-0.137.0), encrypted message delivery is the default for MultiAgentV2 spawn_agent, send_message, and followup_task, so users cannot inspect the exact delegated tasks, follow-up instructions, or intermediate child-to-parent messages as readable text in parent activity rows, child transcript/history, or after resuming a session; the report asks for a user-choosable readable delivery and a structured operational transcript with sender/recipient attribution, states the requested audit trail does not require exposing private reasoning, and notes that #35845 added conditional plaintext-response handling without changing the default, a position the body reaffirms in a source review dated 2026-10-26.
Full record
CLAIM C123
Supported
2026-11-14
GitHub issue #57721 in openai/codex, opened 2026-10-28 by fengzee, records a report that on macOS with Codex desktop app 26.1023.50603 and the Android ChatGPT app connected through Codex/Remote, a new chat created entirely from the phone cannot access the connected Mac's Chrome browser and reports Local browser use is disabled in Work settings although the setting is enabled and both devices use the same account/workspace; sending a message in the same chat from the Mac starts browser access, after which the phone can use it, while chats created on the Mac work when continued from mobile, and restarting both apps and creating a fresh mobile chat does not fix the failure; the report expects mobile-created remote chats to initialize browser access just like desktop-created chats and says initialization failures should not be reported as a disabled setting.
Full record
CLAIM C124
Supported
2026-11-14
GitHub issue #9651 in cline/cline, opened 2026-03-03 by NC17z, records a report that Cline v3.69.0 consistently constructs API requests that exceed the model's maximum context window even when the user message is extremely small, across MiniMax M2.5 and Qwen models on OpenRouter, because it serializes a roughly 4,555-token tool schema into every request, adds roughly 3,000-14,000 tokens of system prompt and text input, and requests the model's full context window as max_output_tokens, exceeding the maximum context by a small margin; the report says the behavior persists even after a full wipe and reinstall of VS Code and Cline, and gives a minimal reproduction of opening a brand-new empty folder and sending a simple message such as hi.
Full record
CLAIM C125
Supported
2026-11-14
GitHub issue #15645 in cline/cline, opened 2026-11-10 by HansChundekad, records a report that in the interactive CLI with auto-approve all on (the default), a plugin beforeTool result of policy autoApprove false is ignored and the tool runs without a prompt, so a plugin cannot require user approval for a risky tool call while auto-approve all is on; with auto-approve all off, the same plugin does get a prompt, even for tools that are normally auto-approved; the report points to requestToolApproval in apps/cli/src/runtime/interactive/approvals.ts, where the auto-approve-all check returns before the request policy is read, and suggests skipping the shortcut when the policy is explicitly false, showing the prompt or denying when headless. Versions cited: Cline CLI 3.0.74 with @cline/core 0.2.0 on macOS arm64.
Full record
CLAIM C126
Supported
2026-11-14
GitHub issue #29915 in openai/codex, opened 2026-06-24 by andrewsohrabi, records a report that permission and approval mode selection is not reliably persisted for new threads or already-loaded existing threads: on a new thread, choosing a non-workspace mode such as Full access falls back to workspace after the thread starts and the first prompt executes, while the picker still looks selectable before submission; on an existing loaded thread, changing approval and permission mode through resume appears to work in the UI, but the live thread keeps the old effective settings; the report proposes two linked fork commits as fix candidates, one routing loaded-thread resume overrides through the validated thread-settings path and one exposing permissions plus activePermissionProfile for thread start, resume, and fork in the stable protocol fixtures, and reports focused local tests passing.
Full record
CLAIM C127
Supported
2026-11-14
GitHub issue #12293 in cline/cline, opened 2026-07-15 by pgalbraith, records a report that Cline 4.0.6 in the JetBrains plugin, using deepseek-chat, entered a loop repeatedly running the same directory listing commands, dir and Get-ChildItem, over 30 times without progress when asked why a CLI executable produced no output; every command returned exit code 0 with no visible output, and Cline could not capture stdout or stderr from any executed process, even simple commands whose output redirection produced no files; Cline kept retrying instead of recognizing the tool limitation, and the user had to repeatedly intervene before Cline gave up with a static analysis answer. System cited: CLion 2026.1.4, Node.js v22.15.0, x64.
Full record
CLAIM C128
Supported
2026-11-15
GitHub issue #65620 in anthropics/claude-code, opened 2026-06-05 by ingo-eichhorst, records a report that since about v2.1.162 (still present on 2.1.165), assistant text blocks are silently dropped from the session transcript whenever the model emits another interleaved thinking block after the text in the same turn, with onset around 2026-06-04; the prose is rendered on screen but appears nowhere in the file, and the two persisted thinking blocks carry distinct signatures; in two sessions analyzed line by line, every turn shaped thinking-text-tool_use kept its text while every turn shaped thinking-thinking-tool_use corresponded to an on-screen prose message absent from the JSONL, and zero persisted text-to-thinking shapes appeared across both sessions; v2.1.158 transcripts from the same machine and model persist every text block. System cited: claude 2.1.165 (also observed on 2.1.162), macOS, model claude-opus-4-8 (1M context).
Full record
CLAIM C129
Supported
2026-11-15
GitHub issue #15360 in cline/cline, opened 2026-10-26 by gregtakacs, records a report that in long tasks the VS Code (Remote-WSL) extension host dies from V8 heap exhaustion, that Cline's panel resets and reopening the same task brings the crash back within 1-2 minutes so the task becomes a crash loop, that there were 33 of these exits in 6 days (18 on 10-21, 6 on 10-23, 9 on 10-26), and that while a turn streams the entire state is re-serialized and pushed to the webview about every 0.4-0.6 s, with one run on 10-26 issuing 1,194 of these pushes in 41 minutes, about 20 GB of JSON serialized, before the host died; the task involved had 5,824 messages and its messages file was 17.3 MB, the same size as each push. The author notes he has not proven whether the serialized strings are retained or are churn the garbage collector cannot keep up with. System cited: Cline 4.1.25, VS Code 1.142.0, Remote-WSL, Windows + WSL2.
Full record
CLAIM C130
Supported
2026-11-15
GitHub issue #15483 in cline/cline, opened 2026-11-01 by Lohnegrim, records a report of two related problems with run_commands, both measured on disk rather than inferred from chat text: (A) a tool call whose input carries an unexpected envelope is rejected with a bare Invalid input that never names the field, while the same validator does name fields for a different error class; (B) context compaction removes the examples that keep the call form correct and locks in the broken form — before the first compaction the session had made 378 run_commands calls, all of them in the correct form, and had never produced the envelope at all, while the compacted prefix retained 10 tool-call examples including 5 run_commands, all five in the broken envelope form, each followed by its retained is_error true result, and zero correct run_commands examples; after the first burst the session produced 17 consecutive rejections and never recovered on its own. System cited: Cline 4.1.26, VS Code 1.144.0, Windows 10.0.26300, provider openai-compatible via vLLM.
Full record
CLAIM C131
Supported
2026-11-15
GitHub issue #10894 in cline/cline, opened 2026-05-19 by DavidHinton-xcpc, records a report that when using VS Code Portable Mode, the MCP server configuration written into settings.json via cline.mcpServers is completely ignored by Cline, that Cline does not read MCP settings from settings.json at all and only reads from a standalone cline_mcp_settings.json file, and that MCP servers defined in settings.json do not appear in Cline's Connected MCP Servers list; the author also reports that in portable mode VS Code stores settings under the portable data directory rather than the standard path, and that the cline_mcp_settings.json file was auto-created at some point. System cited: VS Code 1.120.0 Portable Mode, Cline 3.83.0, Windows 10.
Full record
CLAIM C132
Supported
2026-11-15
GitHub issue #14013 in cline/cline, opened 2026-09-09 by reneehuang1, records a report that an ordinary message fails before streaming starts because an MCP tool's JSON Schema contains regex lookaround unsupported by the selected model provider, and the user does not need to invoke the offending tool because its schema being included in the request is enough to block the entire conversation; the concrete case is resend-mcp 2.19.1, using Zod 4.4.3, whose z.email() fields export pattern values containing negative lookaheads, which are valid JavaScript regex constructs but incompatible with the providers handling the request; Cline displays a misleading model-connection error suggesting a sign-in fix that cannot address the schema rejection; a local tools/list check returned 103 tools, with 18 lookaround-containing patterns across 14 tools; the report notes the failure was not tested with Claude Sonnet 4.5 and lists related issue #11162.
Full record
CLAIM C133
Supported
2026-11-15
GitHub issue #14026 in cline/cline, opened 2026-09-10 by damiandixon, records a report that when running the Cline plugin in CLion, and via the CLI, the server rapidly loops and spams providers.read debug logs, with hundreds of identical providers.read lines generated within milliseconds during startup and during background checks, and startup also triggers bursts of model-catalog Unknown provider id warnings for custom providers; the reporter uses a local-network LLM on an Asus GX10 running vllm, says the setup was working up to about a month ago, and expects the provider configuration read to be cached or debounced. Cline version 4.1.17, JetBrains plugin, CLion 2026.2.2, Ubuntu 22.04.
Full record
CLAIM C134
Supported
2026-11-15
GitHub issue #60058 in openai/codex, opened 2026-11-07 by bradcarroll, records a report that in ChatGPT desktop on Windows, Work mode, app version 26.1103.32239, the new-chat Work composer shows Full access but the resulting conversation's execution context reports workspace-write with restricted network access, and the Full access control is visible only before starting the chat and then disappears, leaving no visible in-thread way to inspect or correct the selected mode; a harmless, explicitly authorized SSH check fails at the first jump-host TCP connection, before SSH authentication, with the SSH child process returning exit 255; fully restarting the ChatGPT desktop app did not change the result, and returning to the same conversation through the web interface produced the same result from its local command runner; the report lists related reports #57671 and #59528 and asks for consolidation with an existing issue if appropriate.
Full record
CLAIM C135
Supported
2026-11-15
GitHub issue #76110 in anthropics/claude-code, opened 2026-07-09 by plittlefield, records a report that Ctrl+clicking a URL printed by Claude Code in its response text opens two duplicate tabs in the browser, both pointing at the same URL, while a plain echo of the same URL in the same terminal opens exactly one tab; the report says Ctrl+clicking a link in Claude Code's output should open exactly one browser tab, and attributes the duplication to Claude Code's own hyperlink rendering, likely the OSC 8 escape sequence it emits around printed URLs, producing two overlapping or adjacent clickable regions for the same visible text. Claude Code version 2.1.205, xfce4-terminal 1.1.3, Linux Mint 22.3.
Full record
CLAIM C136
Supported
2026-11-15
GitHub issue #59159 in zed-industries/zed, opened 2026-06-11 by EV4g, records a report that continuously adding non git-ignored files to a git-synced project, in the reporter's case 200 33 MB .fits files, made Zed start trying to look through the files and use an additional 2 GB of RAM per second, filling up 64 GB of RAM and 64 GB of swap before crashing; memory increased even when the Git window was not open, and after the reporter edited the .gitignore file in a different program Zed stopped requesting 2 GB/s and memory usage went back to normal; the report was filed against Zed 1.6.3 on Pop-OS Cosmic with 64 GB of memory and an AMD HX370 CPU.
Full record
CLAIM C137
Supported
2026-11-15
GitHub issue #47482 in anthropics/claude-code, opened 2026-04-13 by alberduris, records a report that output styles with YAML frontmatter are recognized by Claude Code, shown in the statusline and /config, but the content is not injected into the system prompt, while removing the frontmatter entirely fixes the issue; the reporter found the frontmatter parser appears to recognize and strip the YAML block but fails to inject the remaining content into the system prompt, the style name is correctly extracted yet the instructions after the --- delimiter are lost, and the reporter notes this contradicts the official documentation stating custom output styles are Markdown files with frontmatter and the text that will be added to the system prompt; the report was filed against Claude Code 2.1.104 on macOS.
Full record
CLAIM C138
Supported
2026-11-15
GitHub issue #25015 in openai/codex, opened 2026-05-29 by LostFrxks, records a report that Codex app-server starts additional stdio MCP process stacks when spawning subagents, and after the subagents complete and close_agent is called, some MCP child process trees remain alive under the long-lived codex app-server, so process count and memory usage grow roughly linearly with subagent usage; in a small live repro, 3 trivial subagents took the app-server from 23 descendants and about 203 MiB RSS to 74 descendants and about 3.7 GiB RSS, and after close_agent 9 MCP launcher roots and 33 processes remained; in a larger run, 20 subagents took descendants from 46 to 266 and total RSS from about 2.9 GiB to about 11.7 GiB, with swap usage growing to about 6.4 GiB; the report was filed against codex-cli 0.128.0 on Ubuntu 24.04.4 LTS with 16 GB RAM and 16 GB swap.
Full record
CLAIM C139
Supported
2026-11-15
GitHub issue #49829 in openai/codex, opened 2026-10-01 by xc1427, records a report that opening a Dot-created task in ChatGPT desktop fails with "Failed to resume chat" and "Codex app-server is not available", while the same task opens successfully in ChatGPT web; on a network where direct access to codex-cloud-backend.chatgpt.com times out but an approved proxy works, logs repeatedly show hostKind=durable, failureReason=open_timeout, and WebSocket close code 1006, and installed-code inspection suggests the bundled WebSocket library supplies a custom createConnection without a proxy agent, bypassing Node's environment-aware default agent; the report was filed against ChatGPT desktop 26.928.31416 on macOS with a ChatGPT Pro subscription.
Full record
CLAIM C140
Supported
2026-11-16
GitHub issue #80177 in anthropics/claude-code, opened 2026-07-22 by yetkin-bel, records a report that the iOS Simulator panel in the Claude desktop app cannot start its video stream on macOS 27.0 beta, because the helper process claude-ios-sim crash-loops with an uncaught NSException (SIGABRT) while initializing a CoreImage CIContext, so the panel stays stuck on the "Attach a simulator so Claude can see your app" placeholder and the MCP screenshot action fails; the attach MCP action reports success and the simulator itself is healthy, so the reporter concludes this is host-side in the panel's capture/streaming pipeline; the report was filed against Claude desktop app 1.24012.1 on macOS 27.0 beta with Xcode 26.6.
Full record
CLAIM C141
Supported
2026-11-16
GitHub issue #91763 in anthropics/claude-code, opened 2026-09-03 by sheelcheyne, records a report that after the Claude Desktop auto-updater stages a new MSIX version and force-relaunches the app, every subsequent launch fails with the dialog "Another program is currently using this file" (HRESULT 0x80070020, AppModel-Runtime events 215/208), because the old version's Desktop AppX container is kept alive by a detached git fsmonitor--daemon process that a Claude Code session spawned indirectly; terminating that process destroyed the old container the same second and the next launch succeeded with no reboot and no elevation; the reporter reproduced this three times in 24 hours across three consecutive updates (1.40609.1 to 1.44121.1 to 1.44121.2 to 1.44121.4); the report was filed against Claude Code 2.1.259 bundled with Claude Desktop 1.44121.4.0 on Windows 11.
Full record
CLAIM C142
Supported
2026-11-16
GitHub issue #100377 in anthropics/claude-code, opened 2026-10-08 by moeedahmed, records a report that since about 2026-10-07 21:37 UTC, every claude.ai Project thread sent to run on the reporter's Mac fails to start ("Couldn't start"), with the claude remote-control server log showing "Error: CCR v2 worker registration failed for session cse_...: Request failed with status code 400" for each attempt; it fails on every server, both older 2.1.2xx builds and 2.1.292, while a fresh claude remote-control in a scratch folder on the same machine created its own session and registered its worker normally; the report was filed against macOS with claude remote-control servers run with --capacity 32 --permission-mode bypassPermissions --no-create-session-in-dir.
Full record
CLAIM C143
Supported
2026-11-16
GitHub issue #53065 in anthropics/claude-code, opened 2026-04-25 by AttacktheDPoint-com, records a report that when advisor() is called, the full conversation transcript is forwarded to a second model, and the token usage from both the main executor and the advisor sub-inference are summed in the top-level usage fields, so if auto-compaction logic uses these summed totals the advisor call effectively doubles the apparent context usage, triggering compaction when the main model's actual context is only about 50 percent full; in the reported session the total reported input doubled from 513,354 to 1,028,789 on the advisor turn and the session compacted immediately after, wiping context to about 36K; the report was filed against Claude Code 2.1.111 on Windows 10 Pro with claude-opus-4-6[1m] as the main model and claude-opus-4-7 as the advisor model.
Full record
CLAIM C144
Supported
2026-11-16
GitHub issue #15724 in cline/cline, opened 2026-11-14 by maronsson, records a report that when selecting openrouter/auto or openrouter/auto-code under the OpenRouter provider, Cline reads output_modalities: ["text", "image"] from OpenRouter's model catalog and injects modalities: ["image", "text"] into chat completion requests, and if the user's OpenRouter workspace restricts Auto Router to text-only models, OpenRouter rejects the request with "No endpoints found that support the requested output modalities: image, text."; the report was filed against Cline 4.1.26 with qwen/qwen3.7-flash and reflection/beam-1.0-501b-a23b.
Full record
CLAIM C145
Supported
2026-11-16
GitHub issue #12977 in cline/cline, opened 2026-08-05 by boghyon, records a report that when the model calls an MCP tool via use_mcp_tool with a tool name that the connected server did not advertise, Cline forwards the call to the server unchanged, and the server then rejects it with a bare error such as "Unknown tool: 'get_note'", which is surfaced with no indication of what the valid tool names are, even though Cline already fetched and cached the server's tool catalog at connect time; the report was filed against Cline 4.1.4 (Legacy) with Anthropic Claude Opus 4.8 on a LiteLLM endpoint.
Full record
CLAIM C146
Supported
2026-11-16
GitHub issue #49917 in anthropics/claude-code, opened 2026-04-17 by ARHAEEM, records a report that on an existing install, trying to update or reinstall Claude Desktop fails repeatedly, with the user-facing dialogs showing "Installation failed: AddPackage failed: AddPackage failed with HRESULT 0x80073CF6" and "App installation failed with error message: An internal error occurred with error 0x80073D05."; the reporter's log analysis shows both RemovePackage calls for the existing package failed with 0x80073CFA yet the installer proceeded to AddPackage on top of the non-removed package and claimed success, and every subsequent run failed with AddPackage HRESULT 0x80073CF6 while the package-enumeration step finds nothing to remove; the report was filed against Windows 11 Professional build 10.0.22631 with a previously installed Claude_1.3109.0.0_x64__pzs8sxrjxfjjc.
Full record
CLAIM C147
Supported
2026-11-16
GitHub issue #66523 in zed-industries/zed, opened 2026-11-04 by stefano-scarpone-bluvacanze, records a report that when an external agent session (claude-acp) is resumed with session/load after approving a plan with "clear context and use auto mode", Zed reopens the original session only, and the messages after approval are not shown, because the approval creates a new Claude session and the work done after approval lives in that new session, which Zed does not load; the reporter measured the original session with 100 transcript lines and 12 user messages and the new session with 95 transcript lines and 9 user messages; the report was filed against Zed 1.26.2 and 1.27.0 on Windows 11 Enterprise with claude-acp adapter 0.100.0.
Full record
CLAIM C148
Supported
2026-11-16
GitHub issue #41973 in anthropics/claude-code, opened 2026-04-01 by yuhix-dev, records a report that when Claude Code is run as an MCP server via claude mcp serve, the MCP Agent tool cannot resolve subagents that are visible to the normal Claude CLI: on the same machine claude agents shows general-purpose under user agents, but an MCP Agent call fails with "Agent type 'general-purpose' not found. Available agents:" with the available-agent list appearing empty in MCP server mode; the reporter also tested passing an explicit inline agent definition via --agents into the claude mcp serve process and the MCP Agent call still failed with the same error; the report was filed against Claude Code 2.1.89 on macOS.
Full record
CLAIM C149
Supported
2026-11-17
GitHub issue #13941 in cline/cline, opened 2026-09-08 by spollard, records a report that when read_files fails (for example ENOENT), the tool result payload correctly reports success false with an error string, but every display layer in the CLI renders the call as a success: the non-interactive stream shows a green ok, and the TUI live and session replay paths show an accent-coloured asterisk and 1 lines read from a file that does not exist; the report was filed against cline 3.0.61, also reproduced on cline nightly 3.0.61-nightly.1788783728.
Full record
CLAIM C150
Supported
2026-11-17
GitHub issue #46110 in openai/codex, opened 2026-09-17 by harboriqdev, records a report that Codex 0.155.0-alpha.15 fails before executing any filesystem-restricted command on a native Ubuntu host when /proc/self/mountinfo contains valid nsfs entries created by snapd, with the error "error building bubblewrap command: mountinfo path is not absolute", because daemon_mounts.rs passes the mount root through mount_path(), which rejects any decoded path for which PathBuf is_absolute is false; the report was filed against Ubuntu 26.04.1 with bubblewrap 0.11.1 and snapd providing persisted mount namespaces under /run/snapd/ns.
Full record
CLAIM C151
Supported
2026-11-17
GitHub issue #33493 in openai/codex, opened 2026-07-16 by snrui, records a report that a long-running, image-heavy Codex Desktop thread enters a repeated auto-compaction cycle because the compacted replacement_history keeps historical input_image items including their full inline base64/data-URL payloads; in the reported thread the rollout JSONL was approximately 4.16 GB with 224 compaction records and 92 retained images carrying approximately 26.47 million inline image/base64 characters, leaving only about 8,000 tokens of headroom below the auto-compaction threshold; the report was filed against Codex Desktop 26.707.72221 with embedded CLI codex-cli 0.144.2 on macOS 26.3.
Full record
CLAIM C152
Supported
2026-11-17
GitHub issue #59000 in zed-industries/zed, opened 2026-06-10 by strangemonad, records a report that when a standalone uv project contains an independent inner uv project with its own pyproject.toml, uv.lock and .venv, and the inner project is not a member of the outer one because the outer pyproject.toml has no [tool.uv.workspace] table, uv itself resolves the inner project to its own .venv while Zed selects the outer .venv instead, so Pyright reports Import "cowsay" could not be resolved for packages/inner/main.py; the report was filed against Zed v1.5.5 stable on macOS 26.5.
Full record
CLAIM C153
Supported
2026-11-17
GitHub issue #66782 in zed-industries/zed, opened 2026-11-10 by dob9601, records a report that in Helix mode, copying text with y (yank) goes to internal registers rather than the system clipboard despite the use_system_clipboard always configuration, and that the author believes PR #55455 was an over-correction that caused this; the report was filed against Zed v1.28.0 preview on macOS 27.2 with helix_mode true and vim use_system_clipboard always.
Full record
CLAIM C154
Supported
2026-11-17
GitHub issue #44855 in zed-industries/zed, opened 2025-12-15 by valderman, records a report that when remote editing, Zed is unable to format a buffer that has not been saved to file: opening a new unsaved file, pasting [1,2,3], setting the language to JSON and formatting leaves the file unchanged instead of formatting it to [1, 2, 3], and the report says this is not exclusive to JSON but also happens with at least Python, SQL and Rust; the report was filed against Zed v0.216.0 stable on Linux Wayland Fedora 43.
Full record
CLAIM C155
Supported
2026-11-17
GitHub issue #33624 in openai/codex, opened 2026-07-16 by erikmagnethi, records a report, presented as a publicly reported incident rather than a reproduction, that a GPT-5.6 Sol Ultra-mode subagent running with Full Access expanded a cleanup path incorrectly and recursively deleted nearly all files in the user's Mac home directory; the report states this concerns most contents of the home directory, not independent proof that the entire macOS installation or physical computer was erased, and asks for a hard confirmation and recovery gate for bulk and home-directory deletion even in Full Access, citing a Tech Times report and OpenAI's GPT-5.6 Preview System Card as public evidence.
Full record
CLAIM C156
Supported
2026-11-18
GitHub issue #26984 in openai/codex, opened 2026-06-08 by jacobcxdev, records a report that long-running sessions using stdio MCP servers slowly accumulate open file descriptors in the codex parent process until they hit the per-process soft limit (256 on a default macOS login shell), after which every subsequent open()/spawn() fails with Too many open files (os error 24); the report attributes the leak to orphaned MCP child processes the parent still holds read pipes to, with 186 anonymous PIPE fds observed for a session with about 4-5 configured MCP servers, and was filed against codex-cli 0.137.0.
Full record
CLAIM C157
Supported
2026-11-18
GitHub issue #14367 in cline/cline, opened 2026-09-21 by AbubakarMahmood, records a report that restoring a workspace checkpoint can permanently delete local untracked files protected by an uncommitted .gitignore rule: git reset --hard first reverts .gitignore to the checkpoint state, then git clean -fd deletes files that are newly unignored, with no Recycle Bin or Trash fallback, and the deleted files were never captured in the stash so they are never restored; the report was filed against Cline Desktop v0.0.32 (Core SDK 3.33.x) on Windows 11.
Full record
CLAIM C158
Supported
2026-11-18
GitHub issue #56747 in zed-industries/zed, opened 2026-05-14 by SkeLLLa, records a report that archived ACP sessions keep their MCP server processes alive: after a thread is archived in the Agent Panel, every previously-archived session still has its agent process and all of its MCP child processes running, each holding its own RSS, so memory usage grows monotonically with the number of archived sessions until Zed is fully restarted or each session is explicitly killed; the report was filed against Zed 1.2.3 on Linux.
Full record
CLAIM C159
Supported
2026-11-18
GitHub issue #92016 in anthropics/claude-code, opened 2026-09-04 by DennisPh1977, records a report that Claude Desktop for macOS 1.46388.1 auto-denies the CLI-native SendMessage tool in every Code-tab session, removing subagent resumption entirely because that one tool serves both peer-session messaging and subagent resumption while the desktop-provided replacement only covers peer sessions; the report was filed against Claude Desktop 1.46388.1 with bundled Claude Code 2.1.260.
Full record
CLAIM C160
Supported
2026-11-18
GitHub issue #10549 in cline/cline, opened 2026-05-05 by paviko, records a report that the run_commands tool has a hard 30-second wall-clock timeout and returns a misleading error, Command failed: Command timed out after 30000ms, which the agent interprets as a system-level failure or process crash rather than the tool having stopped waiting, so the agent wastes time and tokens diagnosing a non-problem and may re-run commands that already completed in the background; the report was filed against Cline 0.0.10 (CLI, beta) on Debian 12.
Full record
CLAIM C161
Supported
2026-11-19
GitHub issue #50728 in openai/codex, opened 2026-10-03 by kamiljakubczyk, records a report that Codex native Computer Use repeatedly disrupts manual typing and pasting in Claude Desktop and Opera on macOS, with keystrokes and image pastes duplicating; a read-only CoreGraphics event-tap inspection confirmed two enabled, non-listen-only keyboard taps owned by the same SkyComputerUseService process and attached to Claude, with reported average latencies of approximately 15.7 and 55.6 seconds; the report states this is an observed incident sequence, not a deterministic minimal reproduction, and was filed against Codex desktop app 26.930.31730 on macOS 26.5.
Full record
CLAIM C162
Supported
2026-11-19
GitHub issue #96870 in anthropics/claude-code, opened 2026-09-24 by peterlau03, records a report that on Windows 11 with the MSIX Claude Desktop, every process spawned by a Claude Code session leaks NtFC nonpaged pool whenever it opens a path through a junction or symlink located under %APPDATA% or %LOCALAPPDATA%; on a 16 GB machine this reached 2.7 GB in 9 hours of normal agent work and is freed only by a reboot; the report attributes the root cause to a Windows bug in bindflt.sys, the filter behind MSIX AppData virtualisation; the report was filed against Claude Desktop MSIX 2.7032.0.0 with Claude Code 2.1.280.
Full record
CLAIM C163
Supported
2026-11-19
GitHub issue #105770 in anthropics/claude-code, opened 2026-10-28 by BennyWaitWhat, records a report that Android app sign-in fails on a rooted, unlocked-bootloader device: Google Sign-In succeeds and a Play Integrity token is issued, but the server rejects the sign-in with HTTP 400 from POST /auth/verify_google_mobile, and the app shows "Sign-in couldn't be completed because this app couldn't be verified with Google Play"; on the same device Play Integrity reports BASIC only, with DEVICE and STRONG failing; the report affects both the main Claude Android app and the Claude Duet research-preview app, and was filed for a Pixel 9 Pro on stock Android 17, rooted with Magisk 31.
Full record
CLAIM C164
Supported
2026-11-19
GitHub issue #53408 in anthropics/claude-code, opened 2026-04-26 by sandy9214, records a report that the bundled claude.ai Microsoft 365 MCP connector cannot authenticate personal Microsoft accounts (@hotmail.com, @outlook.com, @live.com): the OAuth flow halts at the Microsoft sign-in page with "You can't sign in here with a personal account. Use your work or school account instead."; the report suggests Anthropic's OAuth app registration uses a work/school-only signInAudience instead of AzureADandPersonalMicrosoftAccount; the report was filed against Claude Code CLI (Opus 4.7) on macOS 25.4.0.
Full record
CLAIM C165
Supported
2026-11-19
GitHub issue #20864 in openai/codex, opened 2026-05-03 by 8th-block, records a report that Codex Desktop App becomes slow/unresponsive when ~/.codex/sessions contains many rollout JSONL files — 109 files totalling 78M in the reporter's case, against 22 entries in session_index.jsonl; the report says Desktop appears to scan/read the entire sessions folder even though session_index.jsonl contains the much smaller Desktop-visible set, and clearing or moving files out of ~/.codex/sessions immediately fixes the lag; the report was filed against Codex Desktop App 26.429.30905 on macOS (Apple Silicon).
Full record