.gitignore

# Python
__pycache__/
*.py[cod]
*$py.class
*.so
.Python
*.egg
*.egg-info/
build/
dist/
wheels/
*.whl

# Virtual environments
.venv/
venv/
env/
ENV/

# uv
.uv/

# Testing / coverage
.pytest_cache/
.coverage
.coverage.*
htmlcov/
coverage.xml
.tox/
.nox/

# Type checkers / linters
.mypy_cache/
.ruff_cache/
.pyre/
.pytype/

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

# OS
.DS_Store
Thumbs.db

# Dev-staged SPA bundle. At wheel-build time pyproject `force-include` copies
# ui/dist -> jentic_one/static; for a local `make start-app` source run the same
# mirror is staged here so the admin surface serves the SPA same-origin. Generated
# from ui/dist, never a source artifact.
/src/jentic_one/static/

# Logs
logs/
.jentic/

# Jentic-managed agent skill written into the CLI dir by `jentic bootstrap` /
# `jentic register` when run from inside the repo (machine-generated, not source).
/cli/AGENTS.md
/cli/.claude/

# Env files
.env
.env.local
.env.*.local

# Config files with real secrets (keep examples and local dev tracked)
config/production.yaml
config/local-sso.yaml
jentic-one.yaml

# Helm: vendored subchart packages and dependency lockfiles. Regenerated by
# `helm dependency build/update` (run via `make helm-deps`).
deploy/helm/**/charts/*.tgz
deploy/helm/**/Chart.lock

# Throwaway local-dev scratch space (never committed).
.scratch/

# Agent task-prompt scratch file (handed to the coding agent; may embed local
# dev credentials — must never be committed).
PROMPT.md

# Claude Code local settings (developer-specific hooks/overrides).
.claude/settings.local.json

# Jentic skill artifacts written into the repo when someone runs
# `jentic bootstrap` / `jentic skill init` from inside the source tree (project
# scope resolves to the cwd). These are generated agent-instruction files, not
# source — keep them out of VCS so a `git add -A` can't commit them. (The
# canonical skill content lives in cli/internal/skillgen/content/.) NB: the repo
# root AGENTS.md is intentionally tracked, so we ignore the generated skills dir
# anywhere plus a generated AGENTS.md only in cli/ (the common launch dir); add
# more paths here if you run the agent from another subdirectory.
**/.claude/skills/
cli/AGENTS.md

# Local SQLite database files.
.data/

# Locally-pinned Go toolchain + module cache (bootstrapped by the CLI dev
# tooling; platform-specific binaries + vendored module cache, never a source
# artifact). Keep out of VCS so `git add -A` can't accidentally commit it.
.toolchain/

# Generated corpus statistics artifact (built by `python -m jentic_one corpus-stats`).
# Global IDF/PMI/NPMI stats for intent-based search — regenerated from the
# operations corpus, never a source artifact (can be tens of MB).
/data/corpus_stats.json

# Local toolchain (Go install for CLI builds)
.toolchain/

# Optional local rules mount. If a `.rules/` directory is present it is read at
# runtime by the arch tests (or point elsewhere via JENTIC_RULES_DIR); it is
# never committed. Without it, the vendored subset under tests/arch is used.
.rules/