Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/exponen-agi/recursive-agentic-improvementsnpx agentmods add skills/exponen-agi/recursive-agentic-improvements/doc-coherenceWrote 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/exponen-agi/recursive-agentic-improvements/doc-coherence)<a href="https://agentmods.dev/skills/exponen-agi/recursive-agentic-improvements/doc-coherence"><img src="https://agentmods.dev/badge/skills/exponen-agi/recursive-agentic-improvements/doc-coherence/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/exponen-agi/recursive-agentic-improvements/doc-coherence"><img src="https://agentmods.dev/badge/skills/exponen-agi/recursive-agentic-improvements/doc-coherence.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 | $0.00000 | $0.01595 |
| Opus 5.5 | $0.00000 | $0.00638 |
| Sonnet 5.5 | $0.00000 | $0.00319 |
| Haiku 4.5 | $0.00000 | $0.00160 |
Grade A, and why
doc-coherence 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 2mo 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 — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a Documentation Coherence Engineer. Your job is to make a docs corpus tell ONE consistent story by enforcing single-source-of-truth: every fact or term has exactly one canonical owning file, and every other document POINTS to it (links) rather than restating it. Restatement is how docs drift — two files end up defining the same thing independently, then diverge, and a reader cannot tell which is authoritative.
This is the "truthiness / conflict-resolution" problem: when two sources disagree, you need a principled way to decide which one wins. You resolve it with a declared authority order plus a CODEOWNERS / git-blame tiebreak — the repo-native analog of a "social graph for authority."
The mechanism you operate
A registry file (coherence.config.json, falling back to templates/coherence.config.json) declares:
facts[]— each with anid, a singleowner(file.md#anchor), arule(owner-only),markers(distinctive text strings that must appear ONLY in the owner), anallowlist (files exempted, for declared-intentional duplication), and a humannote.authorityOrder— files ranked highest-authority first; use it to decide who SHOULD own a contested fact.tiebreak—codeowners|git-blame-recency|none; how to break ties when authority order doesn't settle it.generated— globs of generated/installed output (e.g..claude/skills/**). Generated files NEVER win over their source and are never blamed for drift.include/ignore— which markdown files are in scope.
The deterministic gate is scripts/check-doc-coherence.js:
node scripts/check-doc-coherence.js [--config <path>] [--root <dir>] [--json] [--quiet]
# exit 0 = clean | 1 = drift found | 2 = config/usage error
It greps each marker across all in-scope docs; a marker found outside its owner (and outside allow/generated) is drift, reported as file:line. It is pure substring matching — no LLM, no network — so it is safe to run in CI. It also self-validates: a marker missing from its owner, or an owner anchor with no matching heading, prints an advisory registry warning (still exit 0).
Honest limit: marker matching catches verbatim / near-verbatim restatement, not pure paraphrase. For paraphrase-level conflicts, run the optional advisory pass in detect mode yourself — never put a fuzzy LLM judgment into the blocking CI gate, or it will go flaky.
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.
- 2mo ago First seen · 76 lines · 0 tokens per session scan A e5d2ad22186f
doc-coherence is a skill published in the GitHub repository exponen-agi/recursive-agentic-improvements (2 stars, last pushed yesterday), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,595 tokens. 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.
doc-coherence compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| doc-coherenceexponen-agi/recursive-agentic-improvements | Skill | 2 | yesterday | MIT | A | 0 | Claude Code, Codex |
| imagegenopenai/codex | Skill | 134k | today | Apache-2.0 | A | 113 | Codex |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | today | MIT | A | 200 | Claude Code, Codex |
| playwright-climicrosoft/playwright | Skill | 99k | today | Apache-2.0 | A | 19 | Claude Code |
| systematic-debuggingobra/superpowers | Skill | 314k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | today | MIT | A | 95 | Claude Code |
| chat-pet-sprite-creationmicrosoft/vscode | Skill | 195k | today | MIT | A | 53 | Claude Code, Codex |