.gitignore

# Dependencies
node_modules/

# Build output
dist/
build/

# Environment files
.env
.env.local
.env.*.local

# IDE
.idea/
.vscode/
*.swp
*.swo

# OS files
.DS_Store
Thumbs.db

# Logs
*.log
npm-debug.log*
yarn-debug.log*
yarn-error.log*

# Test coverage
coverage/

# Temporary files
tmp/
temp/
all-models.json

# Claude Code local files
.claude/

# Local claudish profile/config state (per-user, not source)
.claudish.json

# npm lockfile (we use bun.lock)
package-lock.json

# Dev/test files
__tests__/
*.jinja
logs/

# AI session files — the per-session scratch only.
#
# `ai-docs/` itself IS tracked: CLAUDE.md cites durable artifacts that live under
# it, so ignoring the whole tree made those references dangle. Only the
# per-session working directories are ephemeral.
ai-docs/sessions/
ai_docs/sessions/
ai-sessions/
**/ai-sessions/

# Build artifacts
*.tsbuildinfo

# Temp dev files
claude
claude_desktop.flow
# Scratch driver for the resume picker — hardcodes a local absolute path. Written
# unanchored so biome's `vcs.useIgnoreFile` also honours it when run from packages/cli.
demo-picker.tsx

# Debug/analysis artifacts
*.pid
*.mitm
*.offset
analysis_result.txt
content_types.txt
decode_traffic.py
extracted_urls.txt
jetski_service.txt
service_offset.txt
tokens.json
test-results/

# Smoke test results
packages/cli/results/*.json
.worktrees

# Model validation
validation/

# Added by claudeup
.mnemex/
gtd/sessions/
.agents/

# Added by claudeup
# A managed rule says this path should stay tracked in git.
!/.claude/
!/.claude/profiles.json
# A managed rule says this path should stay out of git.
.claude/_profiles/
# Statusline worktree markers are local runtime state.
.claude/.statusline-worktree-*
# Claude cache contents are local generated data and can grow quickly.
.claude/cache/

# Added by claudeup
# Team MCP configuration should be committed so shared tools are discoverable.
!/.mcp.json
/.claude/.coaching
/.claude/gtd