feature-separate-eval-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PATSNAP_KEY | No | Optional PatSnap API key. May be empty if the internal endpoint does not require authentication. | |
| CC_EVAL_DATA_DIR | Yes | Path to the shared ledger data directory. Must be the same absolute path as used by patent-tech-feature-separate. | |
| PATSNAP_API_BASE | Yes | Base URL for the PatSnap API. | |
| PATSNAP_ALLOWED_HOSTS | Yes | Exact hostname allowed for PatSnap API requests (e.g., patent.internal.example). | |
| CC_EVAL_ALLOW_INSECURE_HTTP | No | Set to 'true' to allow HTTP (non-HTTPS) connections to upstream services on trusted internal networks. Defaults to 'false'. | false |
| CLAIM_DECOMPOSITION_JUDGE_TOKEN | Yes | Token for authenticating to the claim decomposition Judge API. This is required. | |
| CLAIM_DECOMPOSITION_JUDGE_API_BASE | Yes | Base URL for the claim decomposition Judge API. | |
| CLAIM_DECOMPOSITION_JUDGE_TOKEN_PREFIX | No | Prefix for the Judge token. Defaults to 'Bearer '. Both 'Bearer ' and 'Bearer' are normalized to a single space. | Bearer |
| CLAIM_DECOMPOSITION_JUDGE_ALLOWED_HOSTS | Yes | Exact hostname allowed for the claim decomposition Judge API (e.g., judge.internal.example). |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| feature_separate_eval_statusA | List recorded decompositions and the next pending record; never uses network. |
| feature_separate_eval_checkA | Check whether one record is evaluable without calling Judge. Patent records query only the CLMS field from the configured PatSnap host and therefore require allow_network=true after user confirmation. Free-text records are checked locally and do not require network permission. |
| feature_separate_eval_runA | Evaluate one decomposition and return its human report and persisted result. This sends claim text, decomposed features, and deterministic checks to the configured Judge. Patent records first send the patent number to PatSnap and request only CLMS. Set allow_network=true only after the user confirms this exact scope. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
Each tool covers a distinct stage of the evaluation workflow: status lists pending records, check performs a pre-flight evaluability test, and run executes the actual evaluation. There is no meaningful overlap or ambiguity between them.
All tool names follow the same feature_separate_eval_ prefix plus a single clear verb: status, check, run. The naming pattern is entirely consistent and predictable.
Three tools is appropriate for this focused evaluation pipeline. Each tool provides a distinct, necessary function, and none are redundant or ornamental.
The tool set covers the full workflow from discovering the next pending record, to verifying evaluability, to running the evaluation and persisting results. No critical missing operation is evident for the server's stated purpose.