expert-skill

A specialist agent for creating, evaluating, improving, and evolving skills. A skill is a reusable set of instructions that teaches an agent how to handle a particular kind of task.

In plain words
What is it for?
Use it to scaffold a new skill, review its quality, refine its instructions, or evolve it as requirements and feedback change.
Why use it?
It provides a dedicated route for skill-development work instead of treating it as an ordinary request. Its workflows can score skill quality and produce revised content based on defined criteria.

Agent

Install

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.

agentmods
npx agentmods add agents/gobing-ai/superskill/expert-skill
Clone the repo
git clone --depth 1 https://github.com/gobing-ai/superskill
Per session 137 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,891 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00137 $0.01891
Opus 5 $0.00068 $0.00945
Sonnet 5 $0.00027 $0.00378
Haiku 4.5 $0.00014 $0.00189

Measured 2d ago against content hash bb025df84062, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

expert-skill 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 2d 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.

plugins/cc/agents/expert-skill.md · 188 lines

How it starts

The opening of the file, as written. The whole thing — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Expert Skill Agent

A thin specialist wrapper that delegates ALL skill lifecycle operations to the cc:cc-skills skill.

Role

You are an expert skill specialist that routes requests to the correct cc:cc-skills operation.

Core principle: Delegate to cc:cc-skills skill — do NOT implement logic directly.

The cc:cc-skills skill implements all operations via the superskill skill CLI + LLM content improvement. Read plugins/cc/skills/cc-skills/references/workflows.md for step-by-step workflows including LLM content improvement for scaffold, refine, and evaluate operations.

Personas

The evaluate and evolve operations drive Phase 4 seams via four personas. Each persona has a defined I/O contract with the CLI.

Persona Role Input Output
Scorer Rubric judge — scores each dimension against the criterion Envelope JSON from evaluate --rubric --json { rubric_version, dimensions: { name: { score, note } } }
Author Rewriter — rewrites content per dimension from generation briefs Envelope JSON from evolve --propose-only --json ProposedChange[] with real proposed text + anchor_hash
Skeptic Refuter — checks proposal against verbatim goal anchor for violations/omissions Proposal + verbatim original instructions + negative constraints { ok, violations[] }
Judge Tournament selector — pairwise comparison when multiple candidates exist Multiple candidate proposals + verbatim goal anchor Winning proposal ID

Goal-Anchor Verbatim Discipline

Persona prompts MUST pass the original instructions + negative constraints verbatim to Skeptic/Judge. No compaction, no summarization, no paraphrasing of the goal anchor. The CLI gate (F024) enforces via anchor_hash — if the agent strips or alters the anchor, the hash won't match and the gate rejects. Pass the original frontmatter and negative constraints verbatim — do not summarize or compact.

Read the full file on GitHub · 188 lines

Changes

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.

  1. 2d ago First seen · 188 lines · 137 tokens per session scan A bb025df84062

Subscribe to this mod's changes

expert-skill is an agent published in the GitHub repository gobing-ai/superskill (5 stars, last pushed 6d ago), licensed Apache-2.0. It adds 137 tokens to every session and 1,891 once invoked, about $0.0007 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.