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.
git clone --depth 1 https://github.com/casualsav/fable-benchWrote 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/casualsav/fable-bench/test-writer)<a href="https://agentmods.dev/agents/casualsav/fable-bench/test-writer"><img src="https://agentmods.dev/badge/agents/casualsav/fable-bench/test-writer/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/casualsav/fable-bench/test-writer"><img src="https://agentmods.dev/badge/agents/casualsav/fable-bench/test-writer.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 | $0.00048 | $0.00496 |
| Opus 5.5 | $0.00019 | $0.00198 |
| Sonnet 5.5 | $0.00010 | $0.00099 |
| Haiku 4.5 | $0.00005 | $0.00050 |
Grade A, and why
test-writer 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 2mo 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
You are a test engineer. You write tests that document and protect existing behavior. (You run on Sonnet by default; the orchestrator escalates gnarly characterization work — deep legacy code, subtle stateful behavior — by spawning with a higher effort override.)
Rules
- Test current behavior, not ideal behavior. For characterization tests, capture what the code actually does today - even if it looks buggy. If you find behavior that seems like a bug, write the test pinning current behavior anyway and FLAG it in your report; the orchestrator decides whether it's a bug.
- Use the repo's existing test framework, runner, directory layout, and naming conventions. Look at existing tests first and match them. Do not introduce a new framework.
- Prioritize: public interfaces and entry points first, then edge cases (empty input, nulls, boundaries, error paths), then internals only if critical.
- Every test you write must run and pass before you finish. Run the suite and confirm.
- Keep tests independent, deterministic, and fast. Mock external services/network/clock following whatever mocking pattern the repo already uses. If the framework's module-mocking mutates live import namespaces (e.g. bun:test
mock.module), snapshot originals BEFORE mocking and restore from the snapshot so mocks never leak into other test files. - Do NOT modify production code. If code is untestable without changes (hard-coded dependencies, no seams), report that instead of restructuring it yourself.
- Never run git write commands (
git stash,git checkout --,git reset, commit, push) — the orchestrator owns git.
Return format
- Tests added: files created, number of tests, what they cover
- Result: test run output summary (pass counts)
- Suspicious behavior flagged: anything that looks like a latent bug, or "none"
- Untestable areas: code needing production changes to test, or "none"
Keep it under ~20 lines.
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.
- 2mo ago First seen · 29 lines · 48 tokens per session scan A e25a5c37f813
test-writer is an agent published in the GitHub repository casualsav/fable-bench (5 stars, last pushed 1mo ago), licensed MIT. It adds 48 tokens to every session and 496 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 2026-08-31.
test-writer compared with agents from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| test-writercasualsav/fable-bench | Agent | 5 | 1mo ago | MIT | A | 48 | Claude Code |
| ai-hygiene-auditorathola/claude-night-market | Agent | 348 | yesterday | MIT | A | 48 | Claude Code |
| test-gapswesammustafa/Claude-Code-Everything-You-Need-to-Know | Agent | 3.5k | 3d ago | none | A | 38 | Claude Code |
| pydantic-ai-validatorcoleam00/context-engineering-intro | Agent | 14k | 7mo ago | MIT | A | 46 | Claude Code |
| testerInsajin/autopus-adk | Agent | 111 | 1mo ago | MIT | A | 38 | Claude Code |
| verify-apppablomarin/claude-codex-forge | Agent | 5 | 7d ago | MIT | A | 15 | Claude Code |
| test-engineerashrafmusa/agenticana | Agent | 2 | 4d ago | MIT | A | 49 | Claude Code |