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/yonatangross/orchestkitWrote 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/rules/yonatangross/orchestkit/adversarial-refutation)<a href="https://agentmods.dev/rules/yonatangross/orchestkit/adversarial-refutation"><img src="https://agentmods.dev/badge/rules/yonatangross/orchestkit/adversarial-refutation/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/rules/yonatangross/orchestkit/adversarial-refutation"><img src="https://agentmods.dev/badge/rules/yonatangross/orchestkit/adversarial-refutation.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 | $0.00035 | $0.01615 |
| Opus 5 | $0.00017 | $0.00807 |
| Sonnet 5 | $0.00007 | $0.00323 |
| Haiku 4.5 | $0.00003 | $0.00161 |
Grade A, and why
adversarial-refutation 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 3d 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 — 114 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adversarial Refutation Engine
Shared protocol for evaluative skills (assess, review-pr, audit-full) to verify
their own decision-bearing findings with a separate, blind refuter — the structural
fix for self-preferential bias. A verifier with skin in the game can't be a fair verifier;
the producer's number, evidence list, and prose ARE its reasoning, so forwarding them
re-introduces the anchoring the pass is meant to remove.
Evidence this matters: the #2217 drift-register audit was ~60% wrong because findings judged themselves. An adversarial pass (this protocol) caught a wrong "KEEP" verdict.
Each skill loads this engine via a thin references/adversarial-refutation.md adapter that
supplies its bindings (which field = "finding", which rubric, where survivors/kills land).
1. Blindness contract
The refuter prompt MAY contain: {file, line, category, raw code slice (PRIMARY artifact), rubric excerpt, sanitized accepted conventions} and a neutral claim naming ONLY the
category + location.
It MUST NOT contain: the producer's identity/subagent_type, the producer's number or
severity label, the producer's description/impact/suggestion prose, the producer's curated
evidence list, sibling findings, the original producer prompt, or "known-weakness" priors.
2. Independent-score-first
The refuter never sees the producer's value. It (a) reads the cited code itself, (b) forms its OWN severity/score band from the rubric, (c) returns an existence verdict. The orchestrator — not the refuter — computes SURVIVES/OVERTURNED by comparing the producer's value against the refuter's independent band. Revise only when the producer value sits OUTSIDE the band, and revise to the near band edge (never the midpoint).
3. Citation-verification gate
Before any KILL/OVERTURN/DOWNGRADE is honored, the orchestrator re-opens the refuter's
cited file:line and confirms it exists and supports the verdict. Unverifiable/hallucinated
citation → downgrade to UPHELD. "No refutation found" or assertion-only (no citation) → UPHELD.
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.
- 3d ago First seen · 114 lines · 35 tokens per session scan A 3ab964a9d6a3
adversarial-refutation is a cursor rule published in the GitHub repository yonatangross/orchestkit (264 stars, last pushed today), licensed MIT. It adds 35 tokens to every session and 1,615 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 2026-09-09.
Other cursor rules, from other repositories
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
cli-error-handling
CLI command error handling patterns.
prefer-direct-imports-over-module-mocks
Prefer extracting a testable core over vi.mock / vi.resetModules when unit tests need to reach production logic entangled with config, env, or singletons.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.