kg_evaluate_retrieval
Phase 5 / E3 — Provenance-aware assessor for a set of chunk_ids returned by kg_search. Returns per-chunk bucket (authored-grounded | extracted-high-conf | extracted-low-conf | no-support), overall distribution, dominant_bucket, and recommend_refusal. Pure metadata read - no LLM cost. Used by the ...
This record as markdown: /tools/io-stablebaseline-sb/kg-evaluate-retrieval.md
What kg_evaluate_retrieval does on Stable Baseline
AI agents call kg_evaluate_retrieval to retrieve information from Stable Baseline without modifying anything. It is typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
| Parameter | Type | Required | Description |
|---|---|---|---|
chunkIds | array | Yes | Array of kg_chunks.id values to assess. |
Parameters from the server's own tool schema.
Why kg_evaluate_retrieval is rated Low
The tool performs a read-only evaluation of chunk metadata retrieved from a search. It has no side effects, modifies nothing, and explicitly states it is a 'pure metadata read'. Misuse potential is minimal as it only returns analytical metadata about previously retrieved content.
From the tool's definition Pure metadata read - no LLM cost. Returns per-chunk bucket...overall distribution, dominant_bucket, and recommend_refusal.
Attacks that exploit this kind of access
The rule that runs kg_evaluate_retrieval safely
PolicyLayer is an MCP gateway: it sits between your AI agents and Stable Baseline, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For kg_evaluate_retrieval, this is the rule to start with:
kg_evaluate_retrieval is read-only, so it stays allowed. Everything else on the server is denied unless you say otherwise.
The button opens the PolicyLayer dashboard: create your workspace, connect Stable Baseline, apply this rule, and every kg_evaluate_retrieval call is checked against it from then on.
Questions about kg_evaluate_retrieval
Phase 5 / E3 — Provenance-aware assessor for a set of chunk_ids returned by kg_search. Returns per-chunk bucket (authored-grounded | extracted-high-conf | extracted-low-conf | no-support), overall distribution, dominant_bucket, and recommend_refusal. Pure metadata read - no LLM cost. Used by the agent's response policy to decide whether to answer confidently, caveat, or refuse. It is categorised as a Read tool in the Stable Baseline MCP Server, which means it retrieves data without modifying state.
kg_evaluate_retrieval accepts 1 parameter: chunkIds. Required: chunkIds. The full parameter table on this page comes from the server's own tool schema.
Register the Stable Baseline MCP server in PolicyLayer and add a rule for kg_evaluate_retrieval: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Stable Baseline. Nothing to install.
kg_evaluate_retrieval is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the kg_evaluate_retrieval rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for kg_evaluate_retrieval. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
kg_evaluate_retrieval is provided by the Stable Baseline MCP server (https://api.stablebaseline.io/functions/v1/cloud-serve/mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on Stable Baseline, and thousands of servers like it.
This server
Across the catalogue