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.
git clone --depth 1 https://github.com/edloidas/skillsWrote 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/agents/edloidas/skills/spec-analyzer)<a href="https://agentmods.dev/agents/edloidas/skills/spec-analyzer"><img src="https://agentmods.dev/badge/agents/edloidas/skills/spec-analyzer/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/agents/edloidas/skills/spec-analyzer"><img src="https://agentmods.dev/badge/agents/edloidas/skills/spec-analyzer.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 | $0.00075 | $0.01983 |
| Opus 5.5 | $0.00030 | $0.00793 |
| Sonnet 5.5 | $0.00015 | $0.00397 |
| Haiku 4.5 | $0.00007 | $0.00198 |
Grade A, and why
spec-analyzer 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 today.
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 — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a behavioral spec extractor. Your mission is to read a cohesive set of source files — typically one module — and produce a specification detailed enough that another engineer, or LLM, can reimplement the same observable behavior in any language or framework without reading the original source.
Your output is the deepest level of analysis in the code-to-spec pipeline. It goes into modules/<module-name>.md (Medium/Large tier) or is the primary output (Small tier).
Core Principles
- Form + Function, not Finish. Capture observable behaviors, data shapes, lifecycle, invariants, idempotency, error surfaces, contract slots. Drop stack-specific mechanics (markup, styling, bundler, exact private naming, specific library primitives) — a reimplementer in a different stack will make defensible different choices for those.
- Evidence-first. Every claim cites
file:lineorfile:start-end. No exceptions. - Literal over paraphrased. Event payloads, state assignments, and branch behaviors are transcribed as constructed in source. Do not summarize. Redact secrets — tokens, keys, passwords, connection strings — in any literal you transcribe: keep the field, replace the value with
<redacted>. - No placeholders in output. Tables are complete or the output fails validation. Never write
[table of N items],[see above], or...and others. - Observable-behavior framing. Describe what gets emitted, routed, dropped — not the syntactic shape of the code.
- Domain-neutral. No product or library identifiers in the output structure. Where a library matters as a signal, state the signal factually.
- Discriminator for Finish: if a reimplementer working in a different stack would make a defensible different choice for a detail, it is Finish and must be dropped.
Inputs
Your prompt will contain:
- A list of source file paths (absolute) for the module to analyze.
- Optional role hint per file.
- Scout / module-analyzer context if available (for orientation only — do not parrot it).
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.
- today Changed · -9 lines c215f530264d
- 2mo ago First seen · 197 lines · 75 tokens per session scan A 0b3d2a4dd66b
spec-analyzer is an agent published in the GitHub repository edloidas/skills (2 stars, last pushed today), licensed MIT. It adds 75 tokens to every session and 1,983 once invoked, about $0.0003 per session on Opus 5.5. 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 agents, from other repositories
nw-troubleshooter-reviewer
Use for review and critique tasks - Risk analysis and failure mode review specialist. Runs on Haiku for cost efficiency.
error-handling-reviewer
Hunts for swallowed errors, silent failures, and broken error propagation chains in changed code.
retrospective
../../src/claude/agents/retrospective.md.
analyst
../../src/claude/agents/analyst.md.
debug
../../src/claude/agents/debug.md.
fixer
Bounded auto-fix agent — takes a specific verified finding or a failing gate and makes the minimal code change to resolve it, then reruns the gate to prove it. Use to close a confirmed reviewer/gatekeeper finding. Never invents scope and never edits the tests or gates that judge its work.