evaluation-protocol

A set of rules for evaluating AI skills through blind tests across different model tiers. It defines how to set pass or fail criteria and how to report results.

In plain words
What is it for?
It is for planning skill tests, judging semantic success rather than exact wording, comparing model tiers, and writing failure analyses with prioritized recommendations.
Why use it?
It keeps evaluations fair by preventing test agents from seeing expected answers and focuses grading on whether the intended result was achieved.

Cursor rule

Install

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.

agentmods
npx agentmods add rules/ai-plugin-marketplace/template/evaluation-protocol
Clone the repo
git clone --depth 1 https://github.com/ai-plugin-marketplace/template
Per session 0 Nothing until a file matches its globs; then the whole rule loads.
When invoked 174 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.00174
Opus 5 $0.00000 $0.00087
Sonnet 5 $0.00000 $0.00035
Haiku 4.5 $0.00000 $0.00017

Measured 2d ago against content hash 583c8200ee22, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

evaluation-protocol 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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

plugins/skill-evaluator/rules/evaluation-protocol.mdc · 28 lines

What it actually says

Evaluation Protocol

Pass/Fail Criteria

  • Define pass/fail criteria for each test case BEFORE executing any test runs
  • Pass criteria should be based on semantic equivalence, not exact string matching
  • A test passes if the output achieves the same goal as the expected outcome

Blind Testing

  • Never expose expected outcomes to test-subject agents
  • Test subjects receive ONLY the skill content and input
  • Any accidental leakage invalidates the test run

Reporting

  • Report results per-tier in a structured table format
  • Include failure analysis with root cause and specific recommendations
  • Order recommendations by expected impact (highest first)
  • The refinement report is the primary deliverable — it must be actionable
Changes

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.

  1. 2d ago First seen · 28 lines · 0 tokens per session scan A 583c8200ee22

Subscribe to this mod's changes

evaluation-protocol is a cursor rule published in the GitHub repository ai-plugin-marketplace/template (10 stars, last pushed 2mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 174 tokens. 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.