agentleFS
samchonsamchon/ttsc/.agents/skills/contracts/SKILL.md

contracts

Skill359 starsChanged 18 days agoMITView on GitHub

Defines the @evidence checklists every maintained ttsc declaration and test answers, and their evidence.config.json wiring. Use when writing or reviewing maintained source or tests, or changing Evidence selection.

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before writing, read the declaration and everything its H2 asks about, such as its callees, callers, tests, and platforms.

    line 20
  2. Distinctive:

    If the code falls short of the question, fix the code, and state any gap that remains instead of claiming the requirement is met.

    line 22
  3. Very specific:

    Write @evidence contracts/<file>.md#<anchor> <reason> after a blank comment line, and @evidenceExclude the same way only for an item that …

    line 23
  4. Distinctive:

    Never narrow a selector or lower a severity to pass the check.

    line 32

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Contracts

Each checklist H2 is a question that every selected declaration answers in its doc comment. A package's evidence.config.json selects the checklists; pnpm evidence checks that answers exist, and review checks that they are true and complete.

ChecklistAsks about
common.mdMethod, design, shortcuts, and documentation of production code
portability.mdNative filesystem, path, and process boundaries
performance.mdComputation cost, reuse, and resource lifetime
testing.mdBehavior, expectations, and cases of every test
e2e.mdBoundaries, shared preparation, and isolation of E2E tests

Answering

  1. Before writing, read the declaration and everything its H2 asks about, such as its callees, callers, tests, and platforms.
  2. Answer every fact the H2 asks for, from what the code actually does. A partial, generic, or untrue answer is a defect, not a pass.
  3. If the code falls short of the question, fix the code, and state any gap that remains instead of claiming the requirement is met.
  4. Write @evidence contracts/<file>.md#<anchor> <reason> after a blank comment line, and @evidenceExclude the same way only for an item that does not apply, never for a whole file.

Selection

  • From a package, pnpm exec evidence list and pnpm exec evidence inspect '<target>' show what a selector reaches.
  • For a declaration the adapter cannot address, keep its native doc comment and place its checklist answers and tags on the nearest supported owner shown by list and inspect.
  • Claims: 3/4 active means a selector matched nothing, without a diagnostic, and Coverage counts checklist sections, not declarations.
  • A Go package of unexported declarations, such as packages/evidence/native, selects no production host, so review covers it. A declare module "@ttsc/lint" augmentation reports typescript-ambient-module until the adapter resolves it.
  • Exclude only generated, copied, or dependency material with verified provenance, and omit a pure re-export barrel whose targets span disjoint claims.
  • Never narrow a selector or lower a severity to pass the check.
  • Enroll tests in the existing Evidence step of build.yml, and record entries the adapter cannot address, such as anonymous callbacks or dynamic registrations, for review.

Editing The Checklists

Tens of thousands of tags cite their paths and H2s, so adding, renaming, or removing an H2 needs an authorized migration of every answer. Keep one chapter per question.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.