janklan Scope: every documentation file in this project (README, CLAUDE.md, docs/, other Markdown) plus code comments that state facts about behaviour. Open the artefact with the list of files reviewed so omissions are visible.
Method: read each documentation file in full and extract every falsifiable claim. Verify the claims in parallel with subagents: read the referenced code verbatim; where reading cannot settle a claim, run the relevant command. Never rely on conversation history or prior knowledge. A claim you cannot settle gets verdict UNVERIFIED with the reason - never a guess. While you're at it, remove all mannered prose. Instead of "a parameter worth varying," the mannered writer produces "a dial worth turning." Instead of "this point still matters," they write "this point earns its keep." The phrases exist to display the writer, not to convey the idea, and readers can tell. That is why mannered prose irritates: it makes the reader work harder so the writer can perform. It is also imprecise. Metaphors drag in connotations the writer did not choose and cannot control. The fix is to say what you mean. When a literal phrase is available, use it.
Findings: false or outdated statements, ambiguities, verbose prose, LLM slop. Every finding must quote the claim, cite its file and line, and cite the evidence (code location or command output). A finding without evidence must not appear.
Identifiers: letter = category (F false, O outdated, A ambiguous, V verbose/slop), number = sequence. Number proposed solutions within their finding (F3.1, F3.2).
Output: a reviewable HTML artefact, not a terminal response. One row per finding: id, location, quoted claim, evidence, proposed solutions. Factual findings first, style after. End with the UNVERIFIED list.