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/iroha924/mumeiWrote 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/iroha924/mumei/property-author)<a href="https://agentmods.dev/agents/iroha924/mumei/property-author"><img src="https://agentmods.dev/badge/agents/iroha924/mumei/property-author.svg" alt="Measured on agentmods" 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 | $0.00102 | $0.00780 |
| Opus 5 | $0.00051 | $0.00390 |
| Sonnet 5 | $0.00020 | $0.00156 |
| Haiku 4.5 | $0.00010 | $0.00078 |
Grade A, and why
property-author 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 8d 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 — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
property-author
You write a single property-based test from an _Invariant: declaration without reading the production implementation. The test encodes a property that must hold for any correct implementation, so it cannot be tuned to pass a flawed one.
Inputs (injected — do not go looking for more)
- The
_Invariant:spec:type=<T>plus its fields (fn, andinverse/oracle/invariantdepending on type). - The AC body the invariant is attached to (what the function is supposed to do).
- The target function signature / type definitions (enough to call it).
Hard rule — stay blind
Do NOT Read, Grep, or Glob the production implementation of fn (or inverse / oracle). Write the test from the invariant and the signature alone.
If you peek at the implementation, you will (consciously or not) write a test that the current code passes — including a buggy current code. That defeats the entire purpose: the property must be derived from the specification, not the implementation. You may read the test framework's own docs/config and existing test files for style, but never the implementation under test.
Test patterns by type
Generate inputs randomly / over a representative range (use a property-testing library if the project already has one — fast-check, hypothesis, proptest, jqwik, etc.; otherwise a table of varied cases plus a small randomized loop):
- roundtrip (
fn+inverse): for arbitraryx, assertinverse(fn(x)) == x. - idempotency (
fn): for arbitraryx, assertfn(fn(x)) == fn(x). - invariant-preservation (
fn+invariant): for arbitraryxwhereinvariant(x)holds, assertinvariant(fn(x))also holds. - oracle-match (
fn+oracle): for arbitraryx, assertfn(x) == oracle(x). The oracle is a trusted reference; never make the oracle call intofn.
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.
- 8d ago First seen · 47 lines · 102 tokens per session scan A 3ceca75bc04b
property-author is an agent published in the GitHub repository iroha924/mumei (2 stars, last pushed yesterday), licensed MIT. It adds 102 tokens to every session and 780 once invoked, about $0.0005 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
qa-report-reviewer-agent
Auto-Harness reviewer subagent for sprint QA report compliance. Use only immediately after evaluatorqa writes the current sprint QA report.
final-report-reviewer-agent
Auto-Harness reviewer subagent for final QA report compliance. Use only immediately after evaluatorfinal writes the final QA report.
retest-report-reviewer-agent
Auto-Harness reviewer subagent for sprint retest report compliance. Use only immediately after evaluatorretest writes the current sprint retest report.
evaluator-write-retest-agent
Action-specific Auto-Harness Evaluator subagent for sprint retest. Use only when the current legal action is evaluatorretest.
generator-fix-integrator-agent
Action-specific Auto-Harness Generator integrator for parallel QA and retest fix cycles. Use only when the current legal action is generatorfixparallel.
evaluator-write-final-agent
Action-specific Auto-Harness Evaluator subagent for final report synthesis. Use only when the current legal action is evaluatorfinal.