← Index

retriEVAL

io.github.hcarrillo001/retrieval-mcpv1.0.0AI & LLMs

LLM evals as MCP tools: score outputs for faithfulness, relevancy, and hallucination.

The current version, v1.0.0, was published to the official MCP registry on 2026-08-24. It is distributed as the Docker image ghcr.io/hcarrillo001/retrieval-mcp:1.0.0, filed under AI & LLMs by this index, and declares 4 environment variables. Removals and unreachable sources are measured across every server this index tracks, contract drift across the servers that answer in consecutive snapshots; the index's current counts put this record in context.

In-path gate · all MCP tools

Using retriEVAL in Claude, Cursor, Gemini CLI, Cline, or Zed?

MCP tool contracts can change remotely with no version bump. The mcpindex gate pins each contract and HOLDs the call when it drifts-before your agent acts. Zero credentials. This is not the package install for this server itself (use Install this server for that).

Install the mcpindex gate (one command)

Rewrites your MCP host config so each server launches behind the gate. Inspect first: curl -fsSL https://mcpindex.ai/install.sh | less

uv tool install mcpindex-gate && mcpindex-config-wire
Auditable install path →Watch it hold a drift →
Trust verdict · v1 advisory · method
NOT YET SCREENEDno verdict on file

Verdict not yet evaluated for this tool. The semantic screen takes adversarial cases first; coverage rolls out as the corpus expands (15/150 labels to graduation). The deterministic conformance probe is built but has not yet run on the public corpus, so a recorded verdict here is REVIEW or UNVERIFIED, never a clearing ALLOW. Until a verdict is recorded, an agent should treat this tool as not-yet-cleared and fall back to its own checks. Method: the eval, four-state verdict, honest limits.

Own this server? Screen its description →

That verdict was true at screening time (snapshot 2026-08-24).

Contracts can change after screening, with no version bump. The gate pins retriEVAL’s tool contracts on first sight and holds any silent change before your agent acts - the check that keeps being true on Tuesday.

See your first HOLD in 2 minutes →

Related: how to trust an MCP server · screen before install · silent contract drift

Environment variables
ANTHROPIC_API_KEY
secret

Judge model API key. Not needed if you set RETRIEVAL_JUDGE_BACKEND=ollama to run the judge locally.

RETRIEVAL_JUDGE_BACKEND

Which judge to score with: anthropic (default), openai, or ollama for a fully local judge.

RETRIEVAL_JUDGE_MODEL

Judge model id, e.g. claude-sonnet-4-6 or qwen2.5:14b for Ollama.

RETRIEVAL_BUDGET_USD

Spend cap for judge calls. Unset means no cap.

MCP quality score · maturity, not trust · methodology
freshness
25
completeness
20
installability
25
documentation
15
stability
10
Alternatives in AI & LLMs
BridgeNode MCP
io.github.bridgenode-ai/bridgenode-mcp

BridgeNode — x402 pay-per-request AI inference. OpenAI-compatible API + MCP server, Solana USDC.

Gemini FAF
one.faf/gemini-faf-mcp

Persistent project context for Google Gemini. Python/FastMCP. IANA-registered .faf format.

costwright — agent budget analysis
io.github.hernaninverso/costwright-mcp

Static worst-case token-budget analysis for LLM-agent workflows + signed budget certificates.