AGENTS.md · git:20261005.46e2fc0 · 2026-10-05 · sha256 1b6833a3afa10079
AGENTS.md git:20261005.46e2fc0A
Immutable. This exact content never changes and is served at /api/v1/blob/1b6833a3afa10079.
This project uses Tink to manage Agent Skills under `.agents/skills/`. ## Testing rules - Never write unit tests after you write code. - Highly prefer E2E tests as the sole testing mechanism. Use them to verify complex features work. At the end of E2E tests, produce a verifiable and repeatable artifact. - If you must test a system in isolation, first write down all the ways it could fail, then write the code. ## git-golden A repository is `git-golden` when all of the following are true: - It is checked out on `main` with a clean working tree. - Local `main` is even with `origin/main`. - Open issues and pull requests are tracked separately; they do not make the checkout unclean. - The latest `ci` run on `main` succeeded. <!-- AI-Native SDLC Router --> ## SDLC Workspace - Read `_system/SDLC.md` for setup, evidence boundaries, and recovery. - Inspect `python3 _system/scripts/sdlc.py status` before creating a run. - Read `stages/<stage-name>/CONTEXT.md` before processing a stage. - Keep factory references in `_shared/` unchanged during feature runs. - Use separate worktrees or clones for code-writing runs. - Stage skills: see the Skills section of the current stage's CONTEXT.md. - Need a specialised skill mid-task? `tink-route --receipt runs/<slug>/skills.jsonl "<what you need>"` prints it on stdout; exit 1 means none fits, so continue without one. <!-- End AI-Native SDLC Router --> <!-- tink:rules begin skillset=build-skillset digest=1f3d3f07b904318d51e96afd8174591e82cabe8ec2da6668bf084aa4d683d4c1 --> Discipline rules for this phase (compiled by tink; do not edit by hand): - principle-build-the-lever: Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks. Build the tool that does it or proves it (codemod, script, generator, or a skill your subagents follow) instead of working by hand. The tool is the artifact a reviewer can rerun. (read: tink mount principle-build-the-lever --json --payload) - unslop: Cut AI tells from any writing. Must always apply. (read: tink mount unslop --json --payload) <!-- tink:rules end -->