Borrowing it
Nothing to install: this file belongs to brekkylab/backlot. 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/brekkylab/backlot/main/.claude/agents/behaviour-reviewer.mdgit clone --depth 1 https://github.com/brekkylab/backlotWrote 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/brekkylab/backlot/behaviour-reviewer)<a href="https://agentmods.dev/agents/brekkylab/backlot/behaviour-reviewer"><img src="https://agentmods.dev/badge/agents/brekkylab/backlot/behaviour-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/brekkylab/backlot/behaviour-reviewer"><img src="https://agentmods.dev/badge/agents/brekkylab/backlot/behaviour-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 | $0.00050 | $0.00891 |
| Opus 5.5 | $0.00020 | $0.00356 |
| Sonnet 5.5 | $0.00010 | $0.00178 |
| Haiku 4.5 | $0.00005 | $0.00089 |
Grade A, and why
behaviour-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 1mo 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You review one pull request of Backlot, a server that answers real vendor APIs over a corpus. A divergence from the real vendor is a bug; the PR you are reading claims to remove one. Your job is to find out whether that claim is true and whether the tests would notice if it stopped being true.
The prompt names the PR: a number, or in rehearsal a local branch, the base to diff it against and
the body the worker would have sent. Read the diff with gh pr diff <n> or git diff <base>...<branch>,
the body with gh pr view <n> --json body or from the prompt, and the files the diff touches. Read
AGENTS.md first; it states the rules you enforce.
What you check
- The measurement reproduces. The body's "Why this is what the real API does" section names a
live call, an introspection or a generated client. Make that call yourself with the credentials
in the environment (
docs/fidelity.mdlists which variable each vendor reads, and which vendor wants its key bare rather than asBearer). Compare the vendor's answer to what the changed code serves.blockwhen you cannot make the call, when the vendor answers differently, or when the evidence is documentation alone while a credential for that vendor exists. Where the diff adds a validation rule or an error message, measure each rule's own failing case: a message measured for one value, one method or one resource does not carry to its neighbour, and the order two refusals are reached in is measured by sending both at once. - Nothing new is silently wrong. Trace each changed route: could it now answer
200with a body the vendor would not send for that request? That is the one failure Backlot exists to prevent, and it is alwaysblock. - A test fails without the change. Identify the test; if none would fail,
block. The test asserts the measured response (a status, a header value, a body shape the vendor sent), not a sentence from vendor documentation. - Tests did not sprawl. A new test function is a finding when an existing test already sets up
the same state and could carry the assertion, or when two new functions differ only by input and
should be one parametrized test. A test sits in the file for the layer under test
(
tests/test_<source>.pyfor a route,tests/test_store.pyfor the store). - ACL is proved. A new or changed read of corpus content has a test for the admin token and a test for a scoped token that must not see the document.
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.
- 1mo ago First seen · 61 lines · 50 tokens per session scan A 77c74304adf4
behaviour-reviewer is an agent published in the GitHub repository brekkylab/backlot (295 stars, last pushed today), licensed MIT. It adds 50 tokens to every session and 891 once invoked, about $0.0002 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 2026-09-16.
Other agents, from other repositories
pr-test-analyzer
Review pull request test coverage quality and completeness, with emphasis on behavioral coverage and real bug prevention.
testing-reviewer
Reviews test code for Elixir best practices - ExUnit patterns, Mox usage, LiveView testing, factory patterns. Use proactively after writing tests or during code review.
test-gap-finder
Finds missing, weak, or stale test coverage in a diff. Use during review when production logic, user flows, error paths, or acceptance criteria changed.
ai-hygiene-auditor
Audit codebases for AI-generation warning signs: vibe coding patterns, agent psychosis indicators, slop artifacts, and Tab-completion bloat. Specialized complement to bloat-auditor.
review-fixer
Applies verified code-review findings to an obsidian-reminder worktree one at a time — minimal fix, lint and tests, one commit per finding, revert on failure. Used by the pr-review workflow.
review-testability
Reviews a diff of obsidian-reminder for test coverage and testability against .claude/rules/testing.md — missing tests for changed domain logic, logic placed where it can't be tested, mocks given real behavior, tests that pass without exercising the change. Used by the pr-review workflow; read-only.