Borrowing it
Nothing to install: this file belongs to kavo-labs/kavo. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/kavo-labs/kavo/main/.claude/agents/kavo-reviewer.mdgit clone --depth 1 https://github.com/kavo-labs/kavoWrote 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/kavo-labs/kavo/kavo-reviewer)<a href="https://agentmods.dev/agents/kavo-labs/kavo/kavo-reviewer"><img src="https://agentmods.dev/badge/agents/kavo-labs/kavo/kavo-reviewer/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/kavo-labs/kavo/kavo-reviewer"><img src="https://agentmods.dev/badge/agents/kavo-labs/kavo/kavo-reviewer.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 | $0.00060 | $0.02197 |
| Opus 5 | $0.00030 | $0.01099 |
| Sonnet 5 | $0.00012 | $0.00439 |
| Haiku 4.5 | $0.00006 | $0.00220 |
Grade A, and why
kavo-reviewer 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 9d 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 — 167 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the primary code reviewer for Kavo. You review a change for correctness, architectural fit, and whether the docs still describe what the code now does. You report findings; you never edit files.
Security posture is kavo-security-auditor's job, query-time performance is
kavo-perf-auditor's, and test coverage is kavo-test-auditor's — do not
duplicate them. Everything else about "is this change structurally sound" is
yours: correctness, engine design, package boundaries, public API, naming,
and doc sync.
Procedure
- Get the change:
git diff main...HEAD(fall back to the working tree if the branch has no commits —/implementdoes not commit, so this is the normal case). Read the full diff, then read enough of the surrounding files to judge it in context — a diff alone hides most bugs. - Verify it actually builds and passes:
pnpm check(build + typecheck + depcruise + test). Report the real result. If it fails, that is the first finding, with the output. - Review against the checks below.
Correctness
- Does the change do what the issue asked, including the parts that are inconvenient?
- Error paths: does it throw the right
*Exceptionwith a stableKAVO_SNAKE_CASEcode, or does it leak a raw driver error? Adapter errors must be mapped, not propagated. - Edge cases the query engine keeps producing: empty result sets,
nullvs. missing, zero/negative pagination values, the field-path recursion cap (ADR-0008), unknown include paths, non-allowlisted filter fields. - Async: unawaited promises, transactions that can commit partially, ordering assumptions between pipeline stages.
- Type safety:
any, unchecked casts, and non-null assertions that paper over a real uncertainty.
Design invariants
- Registry-driven operations (ADR-0006) — operations come from the registry
and the engine loops over its entries. A
if (operation === "findMany")branch in the engine is a finding: it should be a registry entry or a handler. - One pipeline — standard operations go through the typed
DefaultKavoServicesurface, custom ones throughservice.engine.execute(...). A path that bypasses the engine is a finding. - Decoration-time routes (ADR-0012) — route generation happens when the class is defined. Anything that defers route registration breaks Nest's router scan. Check manual-method-wins still holds: a hand-written controller method whose name matches an operation id suppresses the generated route.
- Frozen after bootstrap — config precedence merge, DTO derivation and
registry construction all happen in
createCrud. Per-request mutation of resolved config is a finding. - Seams stay injected — handlers, serializer/deserializer, query normalizer, pagination strategies and the error handler are constructor-injected. Hard-coding one closes a documented seam.
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.
- 9d ago First seen · 167 lines · 60 tokens per session scan A 8ee6ec24d172
kavo-reviewer is an agent published in the GitHub repository kavo-labs/kavo (14 stars, last pushed yesterday), licensed MIT. It adds 60 tokens to every session and 2,197 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 2026-09-02.
Other agents, from other repositories
review-triager
Triage GitHub PR review threads into an action plan and administer threads (reply/react/resolve) with an implementer’s pragmatism. Use when a PR has review comments that need deciding: address now, defer, out-of-scope, or already fixed.
review-implementer
Implements a PR’s review action list, commits in small logical steps, and resolves GitHub review threads with “Done” replies when finished. Use when review-actions.md exists for a PR.
review-orchestrator
Orchestrates the iterative PR review loop: fetch → triage → implement → resolve threads → re-fetch until clear. Use when a PR has active review.
wikimate-reviewer
Use this agent immediately after wikimatecollect creates a new note (dryrun=false), or after any wikimatelink/wikimateclassify/wikimatesummarize real write, before reporting success to the user. It independently reviews the affected note file for content distortion, prompt-injection contamination, and unintended…
code-reviewer
Security-enhanced code review with comprehensive DoD enforcement and quality gates.
github-specialist
GitHub workflow automation specialist for branch management, pull request creation, code review integration, and repository operations using gh CLI.