@heurema/signum
AUse when the user wants contract-first development — define correctness before coding, implement against a contract, audit with multiple models, and package proof artifacts.
Install
agr install @heurema/signum --target claudeWrites 19 files into .claude/skills/, pinned to git-7cfe3537.
- .claude/skills/signum/.gitignore
- .claude/skills/signum/.pre-commit-config.yaml
- .claude/skills/signum/AGENTS.md
- .claude/skills/signum/ARCHITECTURE.md
- .claude/skills/signum/CHANGELOG.md
- .claude/skills/signum/CODE_OF_CONDUCT.md
- .claude/skills/signum/CONTRIBUTING.md
- .claude/skills/signum/DCO.md
- .claude/skills/signum/LICENSE
- .claude/skills/signum/QUICKSTART.md
- .claude/skills/signum/README.md
- .claude/skills/signum/SECURITY.md
- .claude/skills/signum/SKILL.md
- .claude/skills/signum/SUPPORT.md
- .claude/skills/signum/TRADEMARKS.md
- .claude/skills/signum/llms.txt
- .claude/skills/signum/modules.yaml
- .claude/skills/signum/project.glossary.json
- .claude/skills/signum/project.intent.md
Document
name: signum description: Use when the user wants contract-first development — define correctness before coding, implement against a contract, audit with multiple models, and package proof artifacts.
Signum — Evidence-Driven Development Pipeline
Run a CONTRACT → EXECUTE → AUDIT → PACK pipeline for any implementation task.
When to Use
- User asks to "implement X" and you want verifiable correctness
- Task requires multi-model code review
- Changes need proof artifacts for CI/CD gates
- Risk is medium-high and ad hoc coding is insufficient
Pipeline
- CONTRACT — Turn the request into a verifiable contract with acceptance criteria
- EXECUTE — Implement against the contract, not a vague prompt
- AUDIT — Deterministic checks (lint, test, scope) + optional multi-model review
- PACK — Package into a proofpack artifact with verdict (AUTO_OK / HUMAN_REVIEW / AUTO_BLOCK)
Core Rules
- Do not start implementation before a contract exists
- If the contract is vague, stop and improve it
- Check against deterministic criteria, not just model opinion
- Keep active run/pipeline artifacts under
.signum/contracts/<contractId>/ - Treat root
.signum/as the registry/state/archive/compatibility namespace, not the canonical per-run artifact root - If reduced audit coverage on medium/high-risk task → HUMAN_REVIEW, not AUTO_OK
Trustgrade A
- passBody integrity
Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.
- passType matchnot applicable to this artifact type
Whether the artifact is really the kind of thing its metadata claims it is.
- passFreshness
How long since the source repository was last pushed to.
- passPrompt injection
Scans the artifact's own text for instructions aimed at your agent rather than at you.
- passLicense
Whether the source repository declares an SPDX license permissive enough to redistribute.
How the grade is calculated
Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:
- Aevery check passed
- Bone warning
- Ctwo warnings
- Dprompt injection or body integrity failed, or three warnings
- Fone of those failed, and something else is wrong
These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.
Versions
git-7cfe3537bba02026-07-31