bh-rat/steer/references
Claude Code and agent skill by bh-rat/steer
- community
Checking that the SKILL.md file is still in the repository.
Install
npx vskill@latest install bh-rat/steer/referencesSKILL.md
Improving an existing skill
The request is a review, a fix, or a tune-up: the skill already exists and someone is unhappy with it. Work from evidence, not taste.
- Find it.
steer listshows installed skills with a validation
status each; or the user points at a directory.
- Baseline. Run
steer validate <dir>. Fix errors first (spec
violations, broken references, a missing bundled runtime), then warnings, then info. The finding-to-fix map is at the end of writing-the-body.md in this directory.
- Weigh the body. Over budget means the trigger got expensive:
push branch-only sections into references/ behind pointers. Check the description against what users actually say to invoke it; thin or trigger-less descriptions are the most common reason a skill never fires.
- Harvest the lessons. If the skill has learn wired in, run
python3 <skill-dir>/scripts/steer.py learn review: promote the keepers (learn promote <id>), archive the stale (learn forget <id>), both through the same runtime. A learnings.md past 150 lines gets distilled into the body or references, not appended to.
- Replace hand-rolls with components. Grep the body for the
tells: ALL-CAPS step enforcement (flow), probe-the-environment preambles (context), scratch files in /tmp or dot-dirs (store), paste-me-the-key prose (secrets), spawn-and-pray servers (proc). Each swap is steer bundle --with ... plus one copied section, guided by choosing-components.md in this directory.
- Re-verify.
steer validateclean,steer install <dir> --force
where it was installed, then one real run in a fresh context. Read the trace, not just the output: did it load the references it needed, in the order you expected?
- Leave a record. Keep the skill's name and voice; change what
evidence supports. Capture what you changed and why against this skill's own runtime: python3 <building-skills-dir>/scripts/steer.py learn note "<rule>".
The goal is a smaller, sharper skill. Ending with more lines than you started is usually the wrong direction.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/building-skills/references/improving-a-skill.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/bh-rat/steer/references)