backfill-specification
SkillDev toolsIncrementally map a touched legacy test or Plan cluster to approved Specification assertions, drafting a new compact Specification only when none applies
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the backfill-specification skill
What this skill tells your AI
The instructions your AI receives, as published by glowingkitty/openmates in .agents/skills/backfill-specification/SKILL.md and read by ahel’s review.
Workflow
- Run
python3 scripts/specifications.py check-test <path>and inspect the generated registry before creating anything. - Search existing Specifications, neighboring tests, architecture docs, app metadata, REST/CLI/SDK/GUI surfaces, and relevant code to identify the feature.
- If approved assertions apply, add per-test
specification-testmetadata with proof strength and surface, then regenerate the assertion index. - If no approved Specification defines the intended behavior, extract stable truth from the whole feature or Plan cluster, not only one test. Separate historical implementation choices from durable behavior.
- Invoke
define-specification, present the complete compact Specification, and wait for approval before mapping the test or changing implementation behavior. - Update the existing Plan with Specification references and evidence locations; do not replace its discovery, tasks, attempts, handoff, or evidence ledger.
- Run
python3 scripts/specifications.py generateandpython3 scripts/specifications.py check-generatedbefore deploy.
Rules
- Never infer that an existing passing test is approved product truth.
- Search before creating; avoid one permanent Specification per historical Plan.
- Touched legacy tests may not remain
legacy_unmappedat deploy.
Signals
- GitHub stars
- 46
- Forks
- 3
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
backfill-specification- Source
- github.com/glowingkitty/openmates