Agents issues on Windows — page 2
Issues the maintainers labelled both platform:windows and area:agents.
521 issues · 198 open · 323 resolved (62%) · first seen Feb 9, 2026
Is this getting better or worse?
This class of problem is still growing. 260 new reports in the last 90 days vs 171 in the 90 before — +52%. The open backlog peaked at 198 in 2026-10 and sits at 198 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
335 of these reports name the Claude Code build they were running, spanning 139 releases. Heaviest reporters:
- Claude Code v2.1.289
- Claude Code v2.1.258
- Claude Code v2.1.233
- Claude Code v2.1.296
- Claude Code v2.1.123
- Claude Code v2.1.284
- Claude Code v2.1.202
- Claude Code v2.1.92
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 35 days across 323 closures. Of the 323 closures with a recorded reason, 19% were closed as completed and 263 as not-planned or duplicate. The most recent completed fix landed in 2026-10.
Workarounds reported by the community
155 of these issues post a workaround someone says worked. The highest-engagement ones:
- [FEATURE] Add Windows Terminal as a split-pane backend for agent teams (teammateMode)
- …ng an entire Linux subsystem just for pane splitting 3. Use workaround scripts that attempt to bridge the gap via hooks (fragile, unsupported) Proposed Solution Add as a valid backend, with auto-detection support. Detect…
Found in the description of #24384 · still open - Subagent model routing is broken — all mechanisms resolve to parent model (Opus)
- …pe), which the changelog doesn't mention. The full-model-ID workaround noted above likely still applies there.
Found in the thread of #43869 · still open - [BUG] Windows/MSIX 1.28929.0: cross-session messages land in the target's composer but are never submitted — session never responds
- …nd the content to the recipient window directly. 6. Current workaround 1. Write the actual content to a shared file 2. Have a human paste the file path into the recipient window Nothing else got a message through. from t…
Found in the thread of #86069 · still open - [BUG] /model and /effort mutate global settings.json — breaks agents/fleet view; no way to configure per-agent model or effort
- …ents ignoring env values — confirming that even the env var workaround for model configuration was broken for background agents until yesterday. Steps to reproduce 1. Open 2. Launch agent A on Opus — note current model/e…
Found in the description of #66402 · still open - advisor() tool inflates reported input tokens by forwarding full transcript, triggering premature auto-compaction on extended context models
- …ction and destroys the context it was supposed to help with Workaround Setting and/or using a PreCompact hook that blocks auto-compaction: This blocks all auto-compaction and relies on manual . Not ideal but prevents the…
Found in the description of #53065 · still open
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 101–200
Ranked by community engagement (reactions weighted over comments). Listing the top 300 of 521 matches; the statistics above cover all 521.
- Project-scope agent in .claude/agents/ intermittently fails to register at session start (same file, same version — nondeterministic)
- run_in_background: false is ignored when the Agent dispatch includes a name — spawn goes async ("will receive instructions via mailbox")
- Agent tool: uncontrolled recursive subagent spawning — 5 planned agents escalated to 361+ completed background agents, burned full 5h usage quota
- CLAUDE_CODE_SUBAGENT_MODEL=inherit ignores per-call Task model param (contradicts docs)
- Claude Opus 4.7 + ultracode: 50min/37% budget burned, total failure on Chrome Native Messaging integration on Win11
- [BUG] Tool calls silently fail/truncate after interrupting an in-progress subagent-driven execution loop
- [BUG] Agent Teams (in-process): team-lead "active agent" pointer sticks on a teammate after a long/compacted session — lead routes AS the teammate, Agent spawns fail "Teammates cannot spawn other teammates"
- [BUG] Agent Teams: background sub-agents idle for 30+ minutes after work completion; SendMessage ping required to wake them and trigger final report
- Scheduled task ignored repeated user directive — wasted ~210K tokens on prohibited work
- [MODEL] Claude Code stopped delegating to custom subagents and loading skills — regression persists in Opus 4.6
- [BUG] `context: fork` and subagents appear to use a different (lighter) model despite model configuration
- Agent launches background sub-agents without user approval, wasting tokens
- [BUG] Explore subagent should request permission before fetching web content (prompt injection risk)
- [BUG] `disallowedTools: [Write, Edit]` trivially bypassed via Bash tool — agents use sed/awk/redirects to modify files
- [BUG] TodoWrite unrecognized for Subagents
- Bun v1.3.10 panic crash on Windows with multiple local agents
- Gate the idle Ctrl+C "stop all background agents" behind the existing two-press confirmation, or make it opt-out
- [BUG] Desktop Code tab usage chart miscounts tokens: duplicate transcript lines summed, workflow agents skipped, 7d legend shows all-time
- Cross-session messaging (ListAgents/SendMessage) permanently disabled after update, survives reboot/reinstall
- AskUserQuestion hard-freezes CLI tab (Esc/Ctrl+C dead) when options use the preview/"press n to add notes" layout
- Weekend post-mortem, FINAL: the best thing that ever happened to me was running out of Claude usage — four days on Codex delivered what two years on Claude could not
- [BUG] No safeguard before an agent bulk-creates dozens of PRs against a real external repo (91 PRs auto-closed + repo locked me out)
- Subagent `maxTurns` frontmatter is not honored (agent runs past the declared cap)
- Subagent tool calls: PreToolUse denials don't surface to the parent, and hook payload carries the parent's session_id
- [BUG] Hanging find.exe
- Feature request: user interjections should not kill in-flight subagents / background tasks
- [BUG] UTF-8 BOM in agent .md silently prevents agent registration (Agent type not found)
- Subagent ignores explicit instruction to skip test execution
- Claude Code (Opus 4.7) silently narrows user-assigned task scope using self-generated safety heuristic, even when explicit in-session directive and persistent memory forbid it; laundered through external reviewer agent
- 400 'thinking blocks in the latest assistant message cannot be modified' with extended thinking + experimental agent-teams (bridge-session)
- Explore subagent hallucinates "respond text-only" instructions from non-existent prior messages, refuses tools
- Worktree-isolated subagents leak Edit/Write into parent checkout (Windows)
- Sub-agent attempts `exit_plan_mode` as Bash command, fails with Exit code 127
- [BUG] Async subagent loops are silently dropped when the parent has concurrent activity
- [BUG] "Unable to resolve organization UUID" from the RemoteTrigger API
- [BUG] Claude carelessly deletes uncommited work by dropping git stash that failed to pop (commonly repeating pattern)
- [BUG] Task() tool has no timeout — subagent hang leaves orchestrator stuck indefinitely on Windows (no MCP servers required) (refs #28126)
- [BUG] /simplify skill does not specify subagent_type, first agent launch always fails
- PreToolUse/PostToolUse hooks silently disabled in subagents due to CLAUDE_CODE_SIMPLE check in _R()
- [BUG] "Scheduled trigger quota is consumed (trigger_limit_reached) but the Scheduled page is completely empty. Cannot create new trigger. Old trigger ID: trig_01C9F1j37Mak3DP6n9rxh5W7"
- [Bug] Agent incorrectly assumes background tasks are complete
- [Bug] Team member spawn fails intermittently with rate limit error
- [BUG] .claudeignore deny list bypassed by Agent tool sub-process — blocked *.json files were read and modified
- [BUG] Agent memory writes to unpredictable location (project-level vs global)
- Default subagents spawn with ~60k tokens of inherited context; workflow subagents receive the user's latest chat line as their governing instruction
- You've hit your session limit · resets 2:40am (America/Bogota)
- [Bug] Anthropic API Error: False positive safeguard flags on claude-opus-5-5 with reasoning_extraction
- Agent stated unverified work as fact, and hedged about itself to deflect accountability
- [Bug] Sub-agent orchestrators spawn redundant background polling loops instead of awaiting completion notifications
- Subagent transcripts: `message.usage.output_tokens` differs across the block records of one response; the first record carries a partial count
- [FEATURE] Cross-account session messaging between accounts one person owns
- Windows: isolation-worktree safety check rejects valid worktree due to case-sensitive drive-letter comparison
- Fable 5.1 in Claude Code ignores a standing "delegate, don't code" instruction and burns premium usage on grunt work
- [BUG] Subagents can't reach deferred built-in tools (WebFetch, Monitor, DesignSync, …). No ToolSearch to load them, regardless of tools: frontmatter
- [Bug] Cross-session SendMessage to native-Windows silently fails to ingest despite success response
- Stale background-task notifications keep re-firing in long sessions
- Cross-session messaging silently drops messages; 2 app crashes overnight
- Subagent replies delivered to the wrong session when multiple sessions run agents concurrently (Windows)
- Background subagent's final report arrives empty (0 bytes) when 5+ background agents run concurrently
- [BUG] Assigned worktree was an empty/unregistered stub (no .git) — silent fallback to shared parent repo caused a cross-session branch collision
- Subagents on reasoning-heavy open-ended tasks sometimes finish with prose instead of calling the required StructuredOutput tool — request opt-in harness-level enforcement/retry
- Windows: stale-worktree cleanup rm -rf traverses NTFS junctions and deletes data OUTSIDE the worktree (~800 GB data loss)
- [BUG] Main agent using /loop to wait for
- Windows desktop app: session stuck showing 'working' spinner after background agents complete
- [Bug] Transient API 529 "Overloaded" halts an autonomous /goal run permanently — no self-resume, requires manual "continue"
- Workflow tool: args arrives JSON-stringified, not as object; named workflow edits don't hot-reload mid-session
- EEXIST error blocks agent creation when .claude/agents directory already exists
- Multi-lane Workflow agents re-create shared prompt cache per lane (no cross-sibling sharing), causing excessive token usage
- `bgIsolation` guard recovery path is broken: error tells agents to call `EnterWorktree`, which is a deferred tool
- Team in-process mode: process residue on disband + agents don't consume task_assignment
- Write/Edit tools silently succeed (file never persisted) on Windows when target path contains square brackets — dispatched agent context
- Sonnet dispatched via Agent tool fails with "1M context required" error
- Windows: agent-created node_modules junctions from git worktrees cause source-file deletion on worktree remove
- [Bug] Sub-agent recursive delete via typo'd path destroys codebase on NTFS
- claude agents deadlocks when launched with UNC / SMB-mapped working directory (Windows 11)
- Subagents inherit parent prompt cache (~150K tokens), causing plan-mode bleed + self-poisoning hallucinations
- [BUG] Ghost sessions created with unexpected stdin content on Windows when using subagents
- Dispatch beta: dispatcher fails to proactively send_user_message after worker completion (silent until user prompts)
- [Bug] Background Agent produces no deliverable when parent session hits usage limit
- [Bug] Agent scope narrowing causes incomplete audit coverage and missed file verification
- [BUG] Subagents look up agent dependencies as skills instead of using the Agent tool
- [Bug] Agents waste excessive token budget on permission checks without early validation
- Subagents inherit all MCP servers, causing Windows stdio saturation in parallel Task waves
- [BUG] Agent tries to find outside of the project
- [BUG] Sub-agent file writes lost — ephemeral Write tool results not persisted to working directory
- Project agents: only first alphabetical file loaded from .claude/agents/
- [BUG] - Endless loop and excessive tool calls by sub agents contributing to massive spike in token consumption (Max plan user)
- Custom subagents cannot access MCP tools despite docs saying they should
- [BUG] Multiple parallel foreground agents freeze after ~15-30 min, unblock on ctrl+b (move to background)
- [BUG] Explore agent triggers unscoped file access, causing OneDrive to download cloud-only files
- [BUG] Teams not using the ANTHROPIC_MODEL or other env vars.
- [BUG] MCP tools completely inaccessible from `--agent` main thread — breaks orchestrator pattern
- Extension host OOM-crashes within seconds to ~1 min of launching 2+ parallel background Agent subagents (Windows)
- [Bug] Agents consume excessive tokens without producing results
- [Feature Request] Desktop Code tab: invoke custom subagents via @-mention or agent picker
- Make the cross-session send pause ("messaged Claude Desktop sessions 10 times since your user last typed") a configurable setting
- [BUG] Initial agent-type listing never emitted on turn 1 of a session created by plan approval with "clear context"
- Custom subagent silently missing from Agent tool's available list despite valid, sibling-identical frontmatter
- [BUG] Claude Desktop blocks CLI-native SendMessage, breaking subagent continuation
- [FEATURE] Let in-flight subagents resume from their own checkpoint after a usage-limit interruption, instead of restarting from scratch