Model behaviour issues on VS Code — page 3
Issues the maintainers labelled both platform:vscode and area:model.
463 issues · 167 open · 296 resolved (64%) · first seen Feb 26, 2026
Is this getting better or worse?
This class of problem is still growing. 285 new reports in the last 90 days vs 171 in the 90 before — +67%. The open backlog peaked at 167 in 2026-08 and sits at 167 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
355 of these reports name the Claude Code build they were running, spanning 120 releases. Heaviest reporters:
- Claude Code v2.1.170
- Claude Code v2.1.198
- Claude Code v2.1.207
- Claude Code v2.1.210
- Claude Code v2.1.197
- Claude Code v2.1.158
- Claude Code v2.1.239
- Claude Code v2.1.235
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 296 closures. Of the 296 closures with a recorded reason, 10% were closed as completed and 266 as not-planned or duplicate. The most recent completed fix landed in 2026-08.
Workarounds reported by the community
37 of these issues post a workaround someone says worked. The highest-engagement ones:
- [Bug] Anthropic API Error: Model's tool call could not be parsed (retry also failed)
- This issue makes claude code unusable... Quick workaround:
Found in the thread of #62123 · still open - Opus 4.7 thinking summaries not rendered (VSCode extension 2.1.112)
- …ith a capability table), but the suffix variant is missing: Workaround None found that works — the filter strips 4.6 at a later stage, so a byte-level rename of → (length-preserving) has no effect. Users on 4.7[1m] lose…
Found in the description of #49902 · still open - Claude Code changed default model to 1M context without notifying Pro users
- …overnight. I lost hours trying to debug the issue, applying workarounds (environment variables, settings.json), and ultimately had to downgrade to a smaller context model — losing functionality I was already using and pa…
Found in the description of #62199 · still open - [BUG] Server-side experiment (`x-cc-atis`) silently removed Opus 4.8 thinking summaries, and the CLI silently self-updated even with `autoUpdates: false`. No notice, no opt-in, settings silently overridden.
- …de gap, posted the analysis ( and built and shared a public workaround so other affected users could get summaries back: a launcher that injects , the one lever that is not interactivity-gated, plus an extension patch an…
Found in the description of #75607 · still open - [MODEL] Opus 4.8 starts hallucinating results before parallel tasks finish
- …e upstream fix is unknown) 1. is the only durable user-side workaround for both hazards, per longitudinal data in [ 62123 comment 4569840435]( (4.6 clean → 4.7 onset → 4.8 worst, including 1M context). @timmywheels named…
Found in the thread of #63884 · 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 201–300
Ranked by community engagement (reactions weighted over comments). Listing the top 300 of 463 matches; the statistics above cover all 463.
- [BUG] Claude incorrectly claims except A, B: is a SyntaxError in Python
- Opus 4.7: thinking content invisible AND requested work not executed — repeated silent stalls
- [BUG] claude-sonnet-4-6 regression: patch-on-patch reasoning on cascading state, requires explicit prompts to design properly
- [Bug] Model fails to consult loaded memory before creating duplicate entries
- [Bug] Anthropic API Error: Content policy violation on benign educational content
- [Bug] Claude Code generates insecure deployments with exposed credentials and orphaned resources
- [MODEL] Opus 4.7 silently substitutes proxy identifiers in integrity-critical code
- [Bug Report: Regression in code generation quality and increased bug introduction rate
- [Feature Request] Add command to instantly switch AI models (e.g., `/model`)
- Claude Code enters fix loop on mobile image/object-fit issue — 5+ attempts without resolution
- Claude Code fails game developers on specialized platforms — BGA case study
- [Bug] Claude Code fails to resolve CSS image alignment issues
- [Sonnet 4.6] Claude prefers complex solutions over simpler alternatives when multiple options exist
- [BUG] Since updating vscode extension opus only available with 1M context window?
- [Bug] Claude 3.5 Sonnet model performance regression compared to previous versions
- Opus 4.7 (1M): repeated failures in a working repo with documented tooling — 20 logged mistakes across multiple sessions
- [Bug] Model not completing all requested tasks in single response
- [Bug] Model fails to identify root cause in iterative debugging, opts for superficial fixes instead of comprehensive refactoring
- Claude Code: fix-fail-propose-new-fix patching pattern despite explicit project rules forbidding it
- [Bug] Claude lacks access to current time and cannot calculate time intervals
- Claude gave incomplete GitHub issue reports requiring multiple follow-up rounds to get all required fields
- Claude made 3 wrong CSS selector attempts on live site before finding correct Tutor LMS selector
- [BUG] response with japanese during conversaion with korean
- [Bug] Claude generates hallucinated content and cannot provide source links reliably
- [Bug] Claude projecting internal thoughts as user commands, ignoring instructions
- [BUG] Claude Code serves Opus 4.6 despite selecting Opus 4.7
- [Bug] Claude Code generates non-functional code and wastes tokens without executing tasks
- [Bug] Claude model generating incorrect code in extended sessions
- [Bug] Infinite loop in build verification - user can repeatedly trigger verification
- [Bug] Claude exhibits condescending and judgmental tone in responses
- [Bug] Reverse transfer operation incorrectly deletes target records instead of moving source records
- [Bug] Claude Code fails to provide informed guidance for Android Emulator setup
- Log monitoring script scans full log history instead of current run — causes persistent false alarm notifications
- [Bug] Language Mixing and Hallucinated Words in Output
- Claude ignores its own established pattern mid-conversation, doubles token cost
- Systematic model behavior gaps in extended coding sessions -- 55+ documented incidents with actionable improvement suggestions
- [BUG] VSCODE EXTENSION: switching to Sonnet model still gets you opus
- Incorrect Model Version 'Opus 4.1' and System Date Issues
- [BUG] Session titles show model's reasoning preamble instead of generated title
- [BUG] keep getting sonnet for VS even though my setting jason set of opus for deafult
- Agent repeatedly ignores persistent memory instructions (reference docs, commit rules)
- [Bug] Model performance regression: Claude forgetting context from previous sessions
- [Bug] Agent makes assumptions without gathering context or validating input
- [MODEL]
- [BUG] Long-horizon agentic failure: the model investigates a failure instead of searching for the working reference already in the repository
- [Bug] Claude fails to identify and correct repeated errors in task execution
- [Bug] Anthropic API Error: False Positive Detection
- [Bug] Agent generates incorrectly formatted commands
- [Bug] Agent fails to follow user-specified rules or constraints
- [Bug] SafeGuard incorrectly triggered on benign content
- [Bug] Markdown formatting rules not applied in output
- [Bug] Anthropic API Error: Account flagged by safeguards, CVP review pending
- [Bug] Anthropic API Error: Safety filter blocks legitimate code generation requests
- [Bug] Model behavior degradation: Poor task planning and requirement adherence
- [Bug] Models ignoring claude.md configuration directives across versions
- [Bug] CVP Safety Filter False Positive with Active Exemption
- Repeated hard-stop "API Error: Sonnet 5 can't help with this. Start a new session to continue" errors
- [Bug] Tool Integration Flagging Safety Guardrails
- [Bug] Fable model not available in Claude Code
- Agent invents new deliverables instead of following the user's defined workflow step, then repairs its invention through repeated corrections
- I don't have a bug report to analyze in your message. Please provide the actual bug report or issue description you'd like me to create a GitHub issue title for.
- Opus 4.8: thinking blocks emitted with empty text — unexpandable Thinking UI in VSCode
- [Bug] Fable 5 safeguard incorrectly flags benign parser conversation messages
- [Bug] Session summary request fails during website development workflow
- [Bug] Unclear error message for unsupported or restricted operations
- [Bug] Incorrect Content Policy Flag on Safe Development Code
- [Bug] Excessive token consumption on small projects with incomplete analysis
- [Bug] Claude Code enters infinite loop with repeated false positive "fix is in" declarations and excessive processing delays
- [Bug] Model unexpectedly downgraded from Claude 3.5 Sonnet to Claude 3 Opus
- Fable 5's safeguards flagged this message...
- [Bug] Federated learning security research incorrectly flagged as cybersecurity work
- Assistant messages intended for the user are emitted as (encrypted) thinking blocks during tool loops — user sees only 'Thought for Ns', content unrecoverable
- [Bug] Legitimate request incorrectly flagged as security violation
- [MODEL] Any
- Fable 5 session repeatedly auto-downgraded to Opus 4.8 (suspected safety-classifier false positive on normal deploy work)
- [Bug] Model silently downgraded from claude-fable-5 to claude-opus without user action or notification
- False-positive "security threat" flag blocks legitimate defensive security work (forced model downgrade)
- [Feature Request] Improve model selection detection to avoid unintended switches to Opus 4.8
- [Bug] Web scraping tool calls flagged as security violation for public data
- [MODEL] Confidently incorrect answer about Cloudflare Cache Rules evaluation order (first-match vs last-match) — risk of PII exposure via cache misconfiguration
- [Bug] Safeguard False Positive: Routine Code Review Flagged as Sensitive
- [Bug] False positive completion status reported for active task
- Assistant inter-tool narration rendered in Japanese during a Traditional-Chinese-only (zh-Hant) session
- [Bug Report: Fable compatibility issue with workflow requirements]
- [Bug] Model unexpectedly switched to Opus during multi-page HTML refactoring task
- [Bug] Security safeguards incorrectly flag defensive vulnerability analysis as offensive content
- [BUG] Fable 5 safety classifier flags UI/frontend work — trigger appears to be project files in context, not prompts
- [Feature Request] Add security study mode to distinguish legitimate cybersecurity learning from malicious activity
- [BUG] Fable 5 safeguards over-flag routine health/biology work and reroute mid-turn
- [Bug] Model unexpectedly switched from Claude Fable to Claude Opus
- [Bug] Silent model substitution to non-selected model despite Max plan subscription
- [Bug] Silent model downgrade mid-session without user notification
- [Feature Request] Support for authorized web security testing workflows
- [Feature Request] Allow overridable safety filters for academic research on emergent capabilities of sorting algorithms
- [Bug] Incorrect Content Policy Flag on Process Design Translation Task
- [Bug] Safeguard triggered for skeleton medical app scaffold code
- [Bug] Claude Code blocks security hardening on user-owned projects
- [BUG] Model committed with no user request, then quoted a confabulated commit-request message (absent from session JSONL)
- [Bug] Model fabricates user approval and executes destructive operations in plan mode
- [Bug] Inappropriate Content Filter Blocking Valid Research Discussion