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.
npx agentmods add instructions/cyanheads/evals-mcp-server/agents-mdgit clone --depth 1 https://github.com/cyanheads/evals-mcp-serverWrote 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/instructions/cyanheads/evals-mcp-server/agents-md)<a href="https://agentmods.dev/instructions/cyanheads/evals-mcp-server/agents-md"><img src="https://agentmods.dev/badge/instructions/cyanheads/evals-mcp-server/agents-md.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 | $0.06157 | $0.06157 |
| Opus 5 | $0.03078 | $0.03078 |
| Sonnet 5 | $0.01231 | $0.01231 |
| Haiku 4.5 | $0.00616 | $0.00616 |
Grade A, and why
evals-mcp-server AGENTS.md 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 5d 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.
This is a copy
80% identical to obsidian-mcp-server AGENTS.md — 362 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 419 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Developer Protocol
Server: evals-mcp-server
Version: 0.1.3
Framework: @cyanheads/mcp-ts-core ^0.12.3
Engines: Bun ≥1.3.0, Node ≥24.0.0
MCP SDK: @modelcontextprotocol/server ^2.0.0
Zod: ^4.4.3
Read the framework docs first:
node_modules/@cyanheads/mcp-ts-core/CLAUDE.mdcontains the full API reference — builders, Context, error codes, exports, patterns. This file covers server-specific conventions only.
This Server
evals-mcp-server turns an agent into an eval author. It's an authoring-loop server (not a data-source wrapper): the agent produces a verifiable eval record through a draft → review → surgical-revise → submit loop, and the server acts as both scribe (normalize, persist, compile) and adversarial checker (run the record's own grader, reject what doesn't hold up). No external API — the server is the source of truth, backed by plain JSON files under EVALS_DATA_DIR. Full concept + as-built surface: docs/design.md, docs/idea.md, docs/authoring-loop-pattern.md.
Surface (9 tools, 1 resource):
| Primitive | Purpose |
|---|---|
evals_describe_schema |
Required/optional fields + grader options for a task_type (static). |
evals_create_draft |
Create a draft; reflect it back with a review protocol + subagent prompt. |
evals_get_record |
Read a draft or submitted record by id. |
evals_revise_draft |
Surgical set/append/unset patch to a draft by dotted path. |
evals_discard_draft |
Delete a draft. |
evals_run_check |
Run a grader against candidates, PASS/REJECT, decoupled from any record. |
evals_submit_draft |
Committability gate → freeze. |
evals_list_records |
Filter records by status/domain/task_type/tag (summary projection). |
evals_export_records |
Compile submitted records to JSONL/CSV/Inspect/lm-eval under exports/. |
eval://record/{id} |
Resource mirror of evals_get_record. |
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.
- 5d ago First seen · 419 lines · 6,157 tokens per session scan A b6b97739e669
evals-mcp-server AGENTS.md is an instructions file published in the GitHub repository cyanheads/evals-mcp-server (1 stars, last pushed 13d ago), licensed Apache-2.0. It adds 6,157 tokens to every session, about $0.0308 per session on Opus 5. A static security scan graded it A with 0 findings. It is 80% identical to obsidian-mcp-server AGENTS.md, differing in 362 lines, and is treated as a copy.
Other instructions, from other repositories
argus-validation-benchmarks CLAUDE.md
Instructions for pensar-x/argus-validation-benchmarks, covering project overview, what you're building, the goal, success criteria and what is apex?.
evals4j CLAUDE.md
Instructions for dvarahq/evals4j, covering evals4j, layout, build, parity discipline and releasing.
tracelens CLAUDE.md
Instructions for ssf0409/tracelens, covering tracelens - development guide, project overview, ownership boundary, tracelens owns and downstream projects own.
geval copilot-instructions.md
Instructions for geval-labs/geval, covering geval - ai coding instructions, project overview, repository structure, build and test and cli and exit codes.
waku-agent CLAUDE.md
Instructions for ShenSeanChen/waku-agent, covering waku-agent — working conventions, architecture map (file ↔ diagram box), rules and commands.
openbench AGENTS.md
Instructions for minghinmatthewlam/openbench, covering openbench — agent context, local execution context, what openbench is, execution ownership and product goals (the two things we are building toward).