English · Português
House Party Protocol

Method · TIPS.md

Tactical patterns

Short commands to operate the kits without relying on implicit memory. Adjust paths to the emitted directory and read the plan before using --apply. Tips marked new in 2.6.0 need version 2.6.0 or later.

1. Inventory before creating

rg --files | rg '(^|/)(name|termo)'

2. Measure the Git tree without confusing untracked files

git status --short --branch

3. Separate local change from the last commit

git diff --stat && git diff --cached --stat

4. Find the decision, not only the implementation

rg -n -i 'decisão|decision|rejeitad|supersed' docs . --glob '*.md'

5. Run the preflight before the done gate

python frameworks/operator-kit-*/scripts/preflight.py --project .

6. Make the criterion fail before the fix

python -m pytest caminho/do/teste.py::test_caso -q

7. Validate Python syntax in isolation

python -m py_compile caminho/do/script.py

8. Run the artifact's self-test

python caminho/do/script.py --self-test

9. Close multiple criteria with AND

python frameworks/operator-kit-*/scripts/done_gate.py "python -m pytest -q" "python -m py_compile app.py"

10. Declare a partial without painting it green

python frameworks/operator-kit-*/scripts/done_gate.py "python -m pytest -q" --declare-partial "falta validar o destino externo"

11. Install a kit in plan mode first

python installers/kit-forge-*/kit_doctor.py install --kit <kit> --target . --host claude-code

12. Apply the installation only after reading the plan

python installers/kit-forge-*/kit_doctor.py install --kit <kit> --target . --host claude-code --apply

13. Install namespaced skills on Codex

python installers/kit-forge-*/kit_doctor.py install --kit <kit> --target . --host codex --apply

14. Verify the integrity of an emitted kit

python installers/kit-forge-*/kit_doctor.py verify <kit>

15. List registered installations

python installers/kit-forge-*/kit_doctor.py registry

16. Choose a checker from another provider

python multi-session/lane-kit-*/scripts/checker_router.py --maker claude --require

17. See the lanes without editing the registry

python multi-session/lane-kit-*/scripts/lane_board.py --help

18. Count items with an explicit ruler

python frameworks/operator-kit-*/scripts/live_count.py --help

19. Look for common silent errors

rg -n 'except\s+Exception|catch\s*\(|\.catch\(' --glob '*.py' --glob '*.js' --glob '*.ts'

20. Confirm the active GitHub account

gh api user --jq .login

21. Compare the remote before pushing

git ls-remote origin refs/heads/main

22. List MCPs per host

claude mcp list && codex mcp list

23. Generate the catalogue from the tree

python installers/kit-forge-*/tools/catalog_md.py . --write

24. Validate every published skill

find . -type d -name skills -exec python installers/kit-forge-*/tools/skill_lint.py --all {} \;

25. Record an end-to-end run as evidence (new in 2.6.0)

python -m hpp evidence run --id smoke-page --artifact out/report.html --artifact out/smoke.log -- python examples/evidence/smoke_page.py

26. Re-derive a bundle from the files on disk (new in 2.6.0)

python -m hpp evidence verify "$(ls -t .hpp/evidence/smoke-page-*.json | head -1)"

27. Measure a retriever before trusting it (new in 2.6.0)

python -m hpp retrieval eval examples/retrieval/suite.json --retriever-command '["python", "examples/retrieval/keyword_retriever.py"]'

28. Check that every citation resolves (new in 2.6.0)

python -m hpp cite check --text examples/citations/answer.md --context examples/citations/context.json

29. Measure a decider before trusting it (new in 2.6.0)

python -m hpp decide eval examples/typed-decisions/gotcha-family-suite.json --decider-command '["python", "examples/typed-decisions/baseline_decider.py"]'

30. Best-of-N is pass@N: select one, then still verify it (new in 2.6.0)

python multi-session/lane-kit-*/scripts/lane_board.py compete --task T-1 --items T-1a,T-1b --lane lead --model claude-opus-5-5
python multi-session/lane-kit-*/scripts/lane_board.py select --task T-1 --winner T-1a --lane review --model gpt-5.6-sol --reason "smaller diff, same tests"