bh-rat/steer/references

Claude Code and agent skill by bh-rat/steer

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Install

npx vskill@latest install bh-rat/steer/references

SKILL.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.

  1. Find it. steer list shows installed skills with a validation

status each; or the user points at a directory.

  1. 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.

  1. 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.

  1. 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.

  1. 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.

  1. Re-verify. steer validate clean, 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?

  1. 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

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
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/bh-rat/steer/references/badge)](https://verified-skill.com/skills/bh-rat/steer/references)

Something wrong with this skill or its page? Tell us.