Agents issues on WSL
Issues the maintainers labelled both platform:wsl and area:agents.
67 issues · 15 open · 52 resolved (78%) · first seen Mar 4, 2026
Is this getting better or worse?
This class of problem is holding steady. 35 new reports in the last 90 days vs 32 in the 90 before — +9%. The open backlog peaked at 16 in 2026-07 and sits at 15 today.
Derived by counting each matching issue from its creation date until it closed. First-party
computation over the anthropics/claude-code tracker — the trend is not published
anywhere upstream.
Which builds report this
49 of these reports name the Claude Code build they were running, spanning 37 releases. Heaviest reporters:
- Claude Code v2.1.170
- Claude Code v2.1.152
- Claude Code v2.1.119
- Claude Code v2.1.214
- Claude Code v2.1.206
- Claude Code v2.1.201
- Claude Code v2.1.177
- Claude Code v2.1.160
Report volume, not a defect count: an issue can mention a build without that build having caused the failure. Versions come from the version string in the bug-report template, matched against real published tags.
Do these get fixed?
Median time to close is 34 days across 52 closures. Of the 52 closures with a recorded reason, 23% were closed as completed and 40 as not-planned or duplicate. The most recent completed fix landed in 2026-08.
Workarounds reported by the community
20 of these issues post a workaround someone says worked. The highest-engagement ones:
- [BUG] Agent tool isolation: "worktree" has no effect for team agents — agent runs in main repo
- …rktree so the agent works on an isolated copy of the repo." As a workaround, we manually create git worktrees before spawning agents: ...and set in the agent prompt to the worktree path. This works correctly, but the par…
Found in the description of #33045 · resolved - [BUG] Agent Teams: Single teammate spawn creates 10-151 duplicate worker instances, each consuming full context and actively editing files
- …by at least one round-trip latency, not zero. One practical workaround we've been using in an out-of-process coordinator context (external process that calls per agent, tracks subagent files, and refuses to re-spawn if a…
Found in the thread of #55586 · resolved - [BUG] SendMessage silently fails when using agent name — only agent ID works for resuming completed agents
- …schema (using names as documented) consistently fail - The workaround (using agent IDs) contradicts the tool's own documentation - In multi-agent pipelines, this causes wasted agent spawns and redundant work Workaround…
Found in the description of #42999 · resolved - Background subagents do not inherit Write/Edit permission allow rules from settings.json or settings.local.json
- …main-session takeover or foreground subagent fallback. The workaround "run it in foreground" was not discoverable without controlled testing. Ask Either (a) fix propagation so background subagents honor allow rules, or…
Found in the description of #67906 · resolved - [BUG] Agent tool `model` parameter enum rejects 1M-context variants (`opus[1m]`, `claude-opus-4-7[1m]`)
- …ot target it from code. This forces a global settings-level workaround that cannot differ between subagents in the same session (e.g. a persistent Architect needing 1M, while a short-lived one-shot Defect-Fixer is fine w…
Found in the description of #51254 · resolved
Excerpts are quoted from the linked issue so you can judge relevance before opening it. Threads that only ask for a workaround, or report that there is none, are excluded. These are community suggestions, not official guidance from Anthropic — read the full thread before applying one.
Matching issues 1–67
Ranked by community engagement (reactions weighted over comments).
- [BUG] Agent tool isolation: "worktree" has no effect for team agents — agent runs in main repo
- Sonnet model returns 'Rate limit reached' despite 0% usage
- [BUG] Agent Teams: Single teammate spawn creates 10-151 duplicate worker instances, each consuming full context and actively editing files
- [BUG] Agent team subagents do not honor 1-hour prompt cache TTL — main sessions show 100% 1h, every agent session shows 0% 1h
- [BUG] Teammate agent crashes in tmux with error <Box> can't be nested inside <Text> component
- [BUG] TeamCreate/TeamDelete/SendMessage tools not available despite agent teams being enabled
- PreToolUse hook exit code ignored for subagent tool calls
- [BUG] SendMessage silently fails when using agent name — only agent ID works for resuming completed agents
- [BUG] Kernel OOM (not V8 heap) from parallel subagent + per-agent MCP fan-out exhausts host RAM
- [BUG] Anomalous cache read token consumption with agent-based workflows (19.5M tokens for a single fullstack feature)
- [Bug] HTTP 529 overloaded_error is rendered as "Rate limited" and hard-fails parallel sessions/subagents (no backoff, not logged to errors)
- [BUG] Workflow synthesis report contains incomplete and garbage data.
- Background subagents do not inherit Write/Edit permission allow rules from settings.json or settings.local.json
- [BUG] Agent tool `model` parameter enum rejects 1M-context variants (`opus[1m]`, `claude-opus-4-7[1m]`)
- [BUG] --continue resumes from wrong point after parallel Agent execution
- [BUG] Real-time inter-agent messaging for coordinated multi-agent workflows
- [BUG] Task tool with subagent_type "Bash" auto-denies Bash commands when run_in_background: true
- [BUG] Nested subagent (depth-2) completion notifications route to root session and are discarded on 2.1.250 — regression of #75043 item 2
- Credential writes replace symlinks and clobber shared state: concurrent instances sharing .credentials.json cascade each other (and the user) to logged-out
- [BUG] Fullscreen TUI: stale spinner rows + statusline vertical desync under nested/parallel subagent load (WSL2, v2.1.178)
- BUG: Model persistently capitalizes Linux username in paths despite explicit CLAUDE.md rule (regression of #27777)
- [Bug] Main thread can stall after background task completion when Read on output file exceeds token limit
- Feature: Option to separate subagent transcripts from parent session
- Terminated team agents persist in the agent list until session end (no cleanup mechanism) — still occurs on v2.1.186
- [BUG] Workflow tool: runner subprocess leaks memory unbounded (~70 MB/s) → OOM-killed on long multi-agent runs
- [BUG] Async sub-agent completion can rebuild warm prompt cache, quickly burning usage limits in long-running sessions
- [BUG] Agent Teams SendMessage renders teammate message bodies instead of compact summaries
- [BUG] general-purpose subagent returns 500 on Bedrock Opus 4.7 while same profile works on main thread, Plan, Explore
- [Bug] Custom subagent crashes silently when `model: sonnet` — CLI Mo8/createInstance secondary crash hides original error
- Worktree subagents branch from `origin/main` rather than current HEAD — intentional?
- Subagent crashes in retry loop when API returns 400 on image processing
- PreToolUse hooks in subagents receive parent session's transcript_path instead of the subagent's own
- [BUG] 2.1.178: Agent tool ignores team_name parameter (regression from 2.1.177)
- [Bug] Feedback loops cause excessive token consumption with unresponsive background jobs
- [BUG] CLI self-aborts (SIGABRT) at ~6 GB RSS during long background Workflow runs — reproduced on 2.1.160 and 2.1.161 (WSL2)
- Background agents (run_in_background: true) terminate at ~20–30 tool uses with status: completed but mid-thought results, no signal for graceful wrap-up
- [BUG] Opus 4.7 stale background-job reporting + false-completeness framing + checkpoint-stop reflex (multi-hour autonomous session)
- [BUG] Claude Code Issue: Subagent Responses Not Returned to User
- [BUG] MCP tools unavailable in agent-team exec subagents — mcp__*__* returns "No such tool available" despite servers being connected at session level
- tools: frontmatter — misleading error message + silent acceptance of non-documented Agent(a), Agent(b) syntax
- Sandbox mode: files created by background agents via Bash don't persist to real filesystem
- Model hallucinates Human: prefix in assistant response and self-executes fabricated user commands (Agent Teams)
- [Bug] MCP tools not available in spawned sub-agents via Agent tool
- Worktree-isolated agent's worktree is cleaned at turn end while its background children still reference it
- [BUG] Nested background subagent never re-woken by its own children — completion notifications route to the top-level session (2.1.224)
- Desktop/web session rename doesn't propagate to cross-session registry (ListAgents/SendMessage still show derived name)
- EnterWorktree: a stale/empty worktree directory silently resolves to the parent repo, so "isolated" commits land on the parent's checked-out branch
- Workflow subagents receive the session's teammate roster + SendMessage — intended? (works well; docs + an off-switch would help)
- Subagent SendMessage triggers approval popup despite bypassPermissions defaultMode
- Subagent terminated mid-task (budget/context) is reported to parent as "completed" — silent truncation
- [BUG] System reminder causes Claude to create a workflow for a one-line bugfix.
- WebSearch built-in tool returns an empty result envelope inside subagents (2.1.177)
- [BUG] Subagent frontmatter `disallowedTools` silently ignores MCP glob patterns (`mcp__<server>__*`) while `tools` honors them
- [BUG] Subagent dispatch prompt and returned result hidden in main thread — collapsed line cannot be expanded (regression after 2.1.160, still in 2.1.170)
- Sub-agent tool calls/logs incorrectly visible in VS Code extension 2.1.160 (hidden correctly in 2.1.39)
- [BUG] Workflow tool determinism guard false-positives on new Date(/Date.now(/Math.random( substrings inside prompt strings
- [Feature Request] Support mixed execution modes for main thread and agents
- Subagents fail with 400 when using Azure Foundry: effort parameter not supported
- [BUG] Subagent report truncated at ~2.5k chars when flagged 'instruction-shaped pattern'; the suggested SendMessage resend re-runs the agent at full cost and truncates again
- Workflow verify stage fans out 20x past size guideline with no confirmation or cost preview
- Plugin subagents: `model:` frontmatter in a plugin's agents/*.md is ignored (project-level agents honor it)
- Subagent results never returned to the main conversation (Agent tool); result unrecoverable
- isolation:"worktree" subagents: Edit tool silently writes into the PARENT session's worktree (+ git-guard root misattribution, pristine-worktree auto-reap orphans cwd into parent tree)
- Inter-agent wake signals are lossy (edge-triggered, no ack, no idle re-poll) — dispatchers idle forever on completed work
- Feature: show `claude rc` remote control sessions in the agents terminal view
- Subagent misidentifies ephemeral auto-mode <system-reminder> as a prompt injection in Bash tool output
- v2.1.201: turn/message desync with background agents — assistant turn starts before queued message delivery; queued user message silently removed