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/evidence-replay)<a href="https://agentmods.dev/rules/yonatangross/orchestkit/evidence-replay"><img src="https://agentmods.dev/badge/rules/yonatangross/orchestkit/evidence-replay/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/evidence-replay"><img src="https://agentmods.dev/badge/rules/yonatangross/orchestkit/evidence-replay.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 | $0.00040 | $0.01094 |
| Opus 5 | $0.00020 | $0.00547 |
| Sonnet 5 | $0.00008 | $0.00219 |
| Haiku 4.5 | $0.00004 | $0.00109 |
Grade A, and why
evidence-replay 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Evidence Replay Gate
Shared protocol for evaluative pipelines (verify, assess, review-pr): a PRODUCER
finding is inadmissible until the orchestrator has replayed its evidence. A finding that
cites nothing checkable, or whose citation fails replay, must never be averaged into a
score, counted toward a verdict, or printed in a report as if it were verified.
1. Evidence contract
Every producer finding MUST carry exactly one machine-checkable evidence shape:
- Static:
{file, line, quote}wherequoteis a verbatim slice of the cited line(s). - Dynamic:
{command, expected_output}wherecommandis deterministic and side-effect free (read-only git, grep, test run, type-check) andexpected_outputis the exact substring or summary line the producer observed.
A finding with neither shape is not evidence-bearing (see section 4).
2. Deterministic replay before admission
Before the finding enters ANY verdict, score, or report, the orchestrator replays it:
- Static: re-open the cited
file:lineand string-match the quote against the actual content (exact substring match after whitespace normalization, nothing fuzzier). - Dynamic: re-run the cited command and diff its output against
expected_output.
Replay is mechanical. No judgment, no LLM re-interpretation, no "close enough".
3. Mismatch handling: reject, one retry, never drop
On replay mismatch the finding is REJECTED back to the producer together with the replay result (what was actually at that line, or what the command actually printed). The producer gets exactly one retry to re-cite or withdraw. A finding that fails replay is never silently dropped (the rejection and retry outcome are recorded) and never accepted as-is (a wrong citation admitted "because the claim sounds right" is the exact failure this gate exists to stop).
4. Non-replayable findings route to refutation
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 · 98 lines · 40 tokens per session scan A 34d1acb59db8
evidence-replay is a cursor rule published in the GitHub repository yonatangross/orchestkit (264 stars, last pushed today), licensed MIT. It adds 40 tokens to every session and 1,094 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
langsmith_evaluation
Comprehensive guide for LangSmith evaluation with @google/genai, including setup, dataset creation, target functions, evaluators, and running evaluations.
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.