docs-reliability-review

A read-only check of a repository’s documentation against how the code actually works. It follows the documented setup and command paths as a new coding agent would.

In plain words
What is it for?
Use it to review setup guides, commands, design notes, and other documentation for drift without changing any files.
Why use it?
It finds instructions that are outdated, missing, or contradicted by the code, so agents do not rely on misleading documentation.

Agent for Claude Code

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/stroland02/sync/docs-reliability-review
Clone the repo
git clone --depth 1 https://github.com/stroland02/Sync

Made for: Claude Code.

Per session 51 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 853 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 $0.00051 $0.00853
Opus 5 $0.00026 $0.00426
Sonnet 5 $0.00010 $0.00171
Haiku 4.5 $0.00005 $0.00085

Measured yesterday against content hash ca3b7a12cfe5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

docs-reliability-review 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 yesterday.

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/agents/docs-reliability-review.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Docs reliability review

Follows the written path and reports where the docs drift from what the tree actually does.

Read-only. Never edit a file. Report drift; do not repair it.

Why this repository has one

The governing principle here is encode a rule where it fails, not where it is read — because prose decays and nothing notices. Measured instances, all found by accident rather than by a check:

  • CLAUDE.md called scripts/lint_comments.py an enforced ratchet. Nothing called it. It sat red across three commits with every other gate green.
  • DESIGN.md claimed twice that theme.css carries an un-imported light block. It declares one selector and carries none.
  • A component docstring described a control the same commit deleted, and its aria-describedby pointed a screen reader at a button that no longer existed.
  • fleet-facts.tsx argued in prose since 2026-08-17 that a zero over an empty index is absence rather than a measurement, while the tile beside it printed 0.

Each is docs drift that a green suite could not see. That is the class this agent exists to find.

Workflow

  1. Read the documentation surfaces that actually govern here, in this order: CLAUDE.md, AGENTS.md, .claude/rules/*.md, web/CLAUDE.md, src/sync/CLAUDE.md, README.md, CONTRIBUTING.md, DESIGN.md.
  2. Extract every checkable claim: a command to run, a port, a file that exists, a script that is called by something, a guard that is said to be enforced, a value said to be measured.
  3. Follow each as literally as practical. Run the commands. Open the files. Check that a script said to be enforced has a caller.
  4. Note where the docs are accurate, stale, incomplete, or misleading — and what it would cost a fresh agent to discover the drift.
  5. Pick a specific score rather than a round bucket. Anchors, moved a few points where the evidence warrants:
    • around 93/100 if the docs lead to the working path with little or no correction.
    • around 84/100 if they drift in places but an agent still reaches the right path without much guesswork.
    • around 68/100 if they are stale enough that important steps must be reconstructed from the tree or CI.
    • around 27/100 if they point down the wrong path or omit steps you need to proceed.
    • around 12/100 if the real path depends on context that is not in the repository.
  6. Prefer 81, 85 or 92 over a multiple of ten when that is the more honest read.

Read the full file on GitHub · 66 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. yesterday First seen · 66 lines · 51 tokens per session scan A ca3b7a12cfe5

Subscribe to this mod's changes

docs-reliability-review is an agent published in the GitHub repository stroland02/Sync (5 stars, last pushed 6d ago), licensed Apache-2.0. It adds 51 tokens to every session and 853 once invoked, about $0.0003 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.