metis: Skill for Claude Code

.claude/skills/spec-compliance/SKILL.md

spec-compliance is a skill for Claude Code from terrene-foundation/metis. It costs 35 tokens per session (2,609 once invoked), scanned A, original, Apache-2.0.

A code-audit procedure for checking whether an implementation really matches its specification. It inspects the actual classes, functions, and behavior with code parsing and targeted searches rather than trusting that files merely exist.

In plain words
What is it for?
Use it to audit plans, validate generated code, review refactors, and verify that required features and symbols are genuinely implemented.
Why use it?
It catches incomplete or fake implementations that can pass superficial checks, such as a placeholder stream or a required class missing from the expected module.

Skill for Claude Code

Written for Claude Code: installed under .claude/. Also seen: reads .claude/ paths.

This is terrene-foundation/metis's own configuration. It tells Claude Code how to work on metis itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything metis configures →

Reuse

Borrowing it

Nothing to install: this file belongs to terrene-foundation/metis. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/terrene-foundation/metis/main/.claude/skills/spec-compliance/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/terrene-foundation/metis

Made for: Claude Code.

Wrote 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.

agentmods badge for spec-compliance

README.md
[![agentmods](https://agentmods.dev/badge/skills/terrene-foundation/metis/spec-compliance/github.svg)](https://agentmods.dev/skills/terrene-foundation/metis/spec-compliance)
Your own site
<a href="https://agentmods.dev/skills/terrene-foundation/metis/spec-compliance"><img src="https://agentmods.dev/badge/skills/terrene-foundation/metis/spec-compliance/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.

agentmods 80×15 button for spec-compliance

Your own site · 80×15
<a href="https://agentmods.dev/skills/terrene-foundation/metis/spec-compliance"><img src="https://agentmods.dev/badge/skills/terrene-foundation/metis/spec-compliance.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,609 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00035 $0.02609
Opus 5 $0.00017 $0.01305
Sonnet 5 $0.00007 $0.00522
Haiku 4.5 $0.00003 $0.00261

Measured 7d ago against content hash c2d7122895b5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

spec-compliance 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 7d 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.

.claude/skills/spec-compliance/SKILL.md · 228 lines

How it starts

The opening of the file, as written. The whole thing — 228 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Spec Compliance Audit

File existence is NOT compliance. A file wrapper_base.py may exist while none of the spec-required classes inside it do. This skill defines how to verify spec compliance via direct code inspection — AST parsing and targeted greps — instead of trusting that a file at the right path means the right content is there.

When to Use

  • /redteam Step 1 — primary use, every round
  • /codify validation gate before knowledge capture
  • Any "is the plan implemented?" question

The Failure Mode This Prevents

A previous BUILD repo run reported "39/39 PASS, 0 CRITICAL, 0 HIGH" via a convergence-verify.py script. A follow-up deep audit found 27+ CRITICAL findings:

  • StreamingAgent.run_stream() was a fake stream — yielded one synthetic TextDelta from a single inner run_async() call
  • BaseAgentConfig.posture field didn't exist (grep returned 0 hits in entire kaizen/core/)
  • @deprecated decorator was defined in deprecation.py but never imported by base_agent.py
  • client.py was COPIED, not MOVED — both source and destination existed at full size, drifting apart
  • New wrapper modules had ZERO test coverage (grep "from kaizen_agents.wrapper_base" tests/ returned empty)
  • Spec § Security Threats subsections existed only in spec text — no test_<threat> functions

The convergence script existed but it was an existence checker ((path / "__init__.py").exists()), not a compliance checker. The audit must NOT rely on self-reports written by previous rounds.

The Method: Spec → Acceptance Assertions → AST/Grep

For each spec section, write down a literal acceptance assertion table, then verify each row against the actual code. Re-derive every check from scratch each round.

Step 1: Extract Acceptance Assertions From Spec

Read the spec text verbatim. For each promised artifact, write the literal assertion:

Spec promise Acceptance assertion
"StreamingAgent.run_stream() yields TextDelta tokens incrementally" grep def run_stream in src; AST: must yield ≥2 distinct values across the loop, NOT a single yield from a single inner.run_async() call
"BaseAgentConfig has frozen field posture: Posture" grep posture: in BaseAgentConfig dataclass body
"@deprecated decorator applied to 7 extension points" grep @deprecated in base_agent.py — must hit ≥7 distinct methods
"MOVE handler.py from app/legacy/ to app/v2/" source must be deleted OR <50 LOC OR import-and-warn shim

Read the full file on GitHub · 228 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. 7d ago First seen · 228 lines · 35 tokens per session scan A c2d7122895b5

Subscribe to this mod's changes

spec-compliance is a skill published in the GitHub repository terrene-foundation/metis (2 stars, last pushed 4mo ago), licensed Apache-2.0. It adds 35 tokens to every session and 2,609 once invoked, about $0.0002 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 2026-09-03.