Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Silence-view/llm-language --skill refine-skillsgit clone --depth 1 https://github.com/Silence-view/llm-languageWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/silence-view/llm-language/refine-skills)<a href="https://agentmods.dev/skills/silence-view/llm-language/refine-skills"><img src="https://agentmods.dev/badge/skills/silence-view/llm-language/refine-skills/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/silence-view/llm-language/refine-skills"><img src="https://agentmods.dev/badge/skills/silence-view/llm-language/refine-skills.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00115 | $0.02534 |
| Opus 5 | $0.00057 | $0.01267 |
| Sonnet 5 | $0.00023 | $0.00507 |
| Haiku 4.5 | $0.00012 | $0.00253 |
Grade A, and why
refine-skills scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 264 lines — stays where its author put it; the contents beside it link to each section on GitHub.
llm-language:refine-skills — Skill Auditor & Refiner
Overview
A meta-skill that applies llm-language's self-refine methodology at the skill level. Scans user-owned skills, scores them on 10 quality dimensions (see references/skill-quality-rubric.md), and applies non-destructive fixes (catalog in references/skill-fixer-patterns.md).
Core pattern: Generate → Critique → Refine, but for skills instead of prompts.
Complements the Phase 7 light-touch auto-audit in the main llm-language pipeline. Phase 7 detects candidates during any /llm-language invocation; this skill runs the full deep audit on demand.
When to Use
Explicitly invoke via /llm-language:refine-skills when:
- You want a full audit of all your personal skills
- You've been using skills designed for Opus 4.6 and want Opus 4.7 migration
- Phase 7 auto-audit flagged multiple refinement candidates
- You're adopting a new skill pattern (e.g., effort parameter, adaptive thinking) and want to backport
Optional arguments:
/llm-language:refine-skills— audit all user-owned skills/llm-language:refine-skills <skill-name>— audit a specific skill/llm-language:refine-skills --scope=user— only user-level/llm-language:refine-skills --scope=project— only project-level/llm-language:refine-skills rollback <skill-name>— restore from backup
Safety Guarantees
- Scoped to user-owned skills — never touches marketplace-owned plugins or Claude Code built-ins
- Never modifies llm-language itself — recursive self-modification loop is blocked
- Backup before every auto-fix —
.backup-<timestamp>.mdsibling file, kept 30 days - Audit trail — every fix logged to
~/.claude/projects/<project>/memory/skill-refinement-audit.md - Dry-run mode —
--dry-runflag shows what WOULD change without writing - Approval gate — risky fixes (restructure, rewrite) require explicit user approval
- Git-awareness — if skill is in a git repo with uncommitted changes, prompts to stash first
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 264 lines · 115 tokens per session scan A b51d171c595d
refine-skills is a skill published in the GitHub repository Silence-view/llm-language (2 stars, last pushed 3mo ago), licensed MIT. It adds 115 tokens to every session and 2,534 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
github-pr-creation
Creates GitHub Pull Requests with automated validation and task tracking. Use when user wants to create PR, open pull request, submit for review, or check if ready for PR. Analyzes commits, validates task completion, generates Conventional Commits title and description, suggests labels. NOTE - for merging existing…
test-run
Run plugin test suites in this monorepo and report a concise pass/fail summary. Optional plugin slug arg; without arg, runs all plugins under plugins/.
cco-tools
Show what tools actually cost in tokens — learned per-tool averages from observed results, replacing the hardcoded MCP/Agent guesses.
design-system-reference
Style guides and implementation rules for frontend design. Works with design-discovery agent which handles context gathering and VS-based style recommendations. Contains detailed style guides, anti-patterns, and implementation checklists.
magic-compact
Compact this Claude Code session.
boss-briefing
Vault health check — workflow pattern analysis, profile sync, session gap recovery, persona rule proposals.