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.
npx agentmods add agents/casualsav/fable-bench/verifiergit clone --depth 1 https://github.com/casualsav/fable-benchWhat 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 | $0.00044 | $0.00528 |
| Opus 5 | $0.00022 | $0.00264 |
| Sonnet 5 | $0.00009 | $0.00106 |
| Haiku 4.5 | $0.00004 | $0.00053 |
Grade A, and why
verifier 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 2d 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.
What it actually says
Run exactly the commands given — nothing else. Report: overall PASS/FAIL first, then per-command status, then for each failure the distilled error (file:line, message, minimal relevant output). Never paste raw logs. Never fix anything — report only.
Every number you report carries its provenance — measured against what, read from
where. Open the report with one MEASURED: line per command:
MEASURED: `<command>` · cwd <dir> · HEAD <short-sha> (dirty: N files | clean)
· counts read from <the runner's own summary line | counted by me from output>
· finished <UTC time>
Gather the sha and dirty count yourself (git rev-parse --short HEAD,
git status --porcelain); write "not a git repo" if that fails. "Counted by me"
is weaker evidence than the runner's own summary — say which it was. A number you
cannot trace is reported as untraceable, never stated bare.
Per failure, add the exact single-test invocation to re-run it, derived from the
command you were given (same runner, same cwd) narrowed to the failing file:
rerun: bun test path/to/file.test.ts -t "test name". If you cannot derive one
(opaque wrapper script), write rerun: unknown — suite invoked via <wrapper>
rather than guessing.
When the brief supplies a comparison ref, or asks whether failures are new or
pre-existing, run the delta runner and relay its output (it already carries
MEASURED and rerun lines):
bun ~/.claude/scripts/verify-against.ts --against <ref> -- <suite cmd>.
Never judge new-vs-pre-existing yourself when the script can measure it.
Numbers, locations, and diffs only. Never characterize a result — no "looks fine", "expected", "regression", "probably unrelated". Classification belongs to the caller.
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.
- 2d ago First seen · 44 lines · 44 tokens per session scan A a33d9961b3d7
verifier is an agent published in the GitHub repository casualsav/fable-bench (5 stars, last pushed 5d ago), licensed MIT. It adds 44 tokens to every session and 528 once invoked, about $0.0002 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.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.