# Greybook

> Current knowledge for people and coding agents. When an agent gets stuck,
> Greybook has what is working right now — the fix, the versions it applies
> to, and what already failed — for Claude Code, Codex, Cursor, Gemini CLI
> and OpenCode.

Every issue page carries an explicit status, the current fix, approaches that
failed, and a reproduction matrix across agent, version, operating system, and
framework. Every count on the site is a count of individual reports that can
be expanded into its rows; member and anonymous confirmations are tallied
separately. Recommendations state their support level (strongly supported,
weakly supported, disputed, outdated) rather than implying certainty. Pages
seeded by editors from public sources are stamped as such and cite those
sources.

When citing a page, prefer quoting its status, affected version range, and
last-verified date together — agent behaviour is version-specific and a fix
that is correct for one release is often wrong for the next.

## Agents covered

- Claude Code (Anthropic), latest 2.1.220: https://www.greybook.io/agents/claude-code
- Codex (OpenAI), latest 0.146.0: https://www.greybook.io/agents/codex
- Cursor (Anysphere), latest 3.12: https://www.greybook.io/agents/cursor
- Gemini CLI (Google), latest 0.53.1: https://www.greybook.io/agents/gemini-cli
- OpenCode (Anomaly), latest 1.18.11: https://www.greybook.io/agents/opencode

## Issues with a verified fix

- [Claude Code keeps rewriting package-lock.json](https://www.greybook.io/problems/claude-code-rewrites-package-lock) — Claude Code, affects 2.1.217 – 2.1.220, 2 member confirmations, last verified 2026-08-03
- [API Error: 400 due to tool use concurrency issues. Run /rewind to recover the conversation.](https://www.greybook.io/problems/claude-code-api-error-400-tool-use-concurrency) — Claude Code, affects reported across 2.1.216 – 2.1.220; not fixed as of 2.1.220, 1 member confirmations, last verified 2026-08-02
- [ProviderModelNotFoundError on a model the provider clearly serves](https://www.greybook.io/problems/opencode-provider-model-not-found-error) — OpenCode, affects long-standing; reproduced on 1.18.4 and 1.18.11, 1 member confirmations, last verified 2026-08-01
- [API Error: Claude's response exceeded the 32000 output token maximum](https://www.greybook.io/problems/claude-code-response-exceeded-32000-output-token-maximum) — Claude Code, affects 2.1.216 – 2.1.220 (a limit, not a regression), 2 member confirmations, last verified 2026-08-01
- [MCP server keeps disconnecting mid-session](https://www.greybook.io/problems/mcp-server-keeps-disconnecting) — Claude Code, affects 2.1.217 – 2.1.220 (client side); varies by server, 2 member confirmations, last verified 2026-08-01
- [OpenCode 1.17.10 crashes with a Bun segmentation fault on Windows](https://www.greybook.io/problems/opencode-bun-segfault-windows-1-17-10) — OpenCode, affects 1.17.10 on Windows; 1.17.9 unaffected; resolved in the 1.18 line, 0 member confirmations, last verified 2026-07-14
- [Codex claims tests passed when the build failed](https://www.greybook.io/problems/codex-claims-tests-passed) — Codex, affects 0.144.0 – 0.146.0, 2 member confirmations, last verified 2026-07-31
- [Agents sprinkle 'use client' and break server component boundaries](https://www.greybook.io/problems/nextjs-agent-breaks-server-components) — all agents, affects All current agents; App Router projects (Next 14–16), 2 member confirmations, last verified 2026-08-01
- [PreToolUse hooks never fire under PowerShell profiles on Windows](https://www.greybook.io/problems/claude-code-hooks-not-firing-windows) — Claude Code, affects 2.1.217 – 2.1.218; fixed in 2.1.219, 2 member confirmations, last verified 2026-07-14

## Workflows

- [Use Claude to review Codex output, step by step](https://www.greybook.io/recipes/use-claude-to-review-codex-output) — 2 confirmed runs. Codex is fast and overshoots; Claude is a pedantic reviewer. Chaining them is the exact workflow that saves me hours: Codex writes on a branch, Claude reviews the diff against a fixed checklist, and nothing merges until the checklist passes.
- [Safe parallel agents with Git worktrees](https://www.greybook.io/recipes/safe-parallel-agents-git-worktrees) — 4 confirmed runs. Run three or four agents against the same repo without them stomping each other's changes. One worktree per agent, one branch per worktree, and a cleanup routine that keeps the repo sane.
- [Prevent coding agents from reading production secrets](https://www.greybook.io/recipes/prevent-agents-reading-production-secrets) — 2 confirmed runs. Agents read everything they can reach, including the .env you forgot about. Layered defense: get secrets out of the tree, deny-list what must stay, and plant a canary so you find out when the layers fail.
- [Wire Claude Code into the Xcode build-and-test loop](https://www.greybook.io/recipes/xcode-claude-code-build-test-loop) — 1 confirmed runs. Claude Code can't see Xcode's build errors unless you hand them over. XcodeBuild MCP plus a formatter that strips the noise gives the agent a real edit-build-test loop for SwiftUI work — the difference between guessing and iterating.

## Configuration files

- [The minimal AGENTS.md that actually works](https://www.greybook.io/configs/minimal-agents-md) — Fifteen lines, ~180 tokens, and it outperforms the 400-line files it replaced. Everything an agent needs on every task; nothing it needs only sometimes. The controversial part is what's missing.
- [Production AGENTS.md for SwiftUI apps](https://www.greybook.io/configs/swiftui-production-agents-md) — The AGENTS.md running my shipping iOS app. Simulator names, the exact xcodebuild invocations, SwiftUI conventions the agent kept violating until they were written down, and a definition of done that includes a green test run.

## Comparisons

- [Codex vs Claude Code for large monorepos](https://www.greybook.io/comparisons/codex-vs-claude-code-large-monorepos) — 200k+ LOC monorepos, strong support

## Machine-readable

- Sitemap: https://www.greybook.io/sitemap.xml
- Feed: https://www.greybook.io/rss.xml
- Search suggestions: https://www.greybook.io/api/search/suggest?q=QUERY