Loading ToolBench

Reading the current corpus…

matterpilot MCP server · D · ToolBench

matterpilot

Read-only MCP server exposing matter, notice, case, and document state for AI clients. Scoped to workspace; supports semantic search over indexed documents, hearing schedules, and case timelines.

2026-07-28+STDIO5 tools

57/100

Definition Quality78

Protocol Readiness35

Supportability35

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

MatterPilot exposes 5 read-only tools with consistent naming (verb_noun pattern: list_, get_, search_) and reasonable descriptions (avg ~150 chars). All tools have input schemas with types and enums where appropriate. However, parameter descriptions are sparse or missing for several fields (e.g., 'limit' in list_matters lacks guidance on why 50 is default). Output schemas are not documented, responses are JSON-stringified but field structure is not declared. Error handling is minimal; no recovery guidance for common failures (e.g., 'matter not found', 'RAG service unavailable'). Security is sound (read-only, no secrets in params, workspace-scoped), but composition could be tighter, search_matter_rag requires OPENAI_API_KEY server-side but this dependency is not surfaced in error messages or tool metadata.

Tools (5)

get_case_notice_timelineread onlysource verified78/100

Fetch every notice, hearing, and follow-up task on a single bankruptcy case. Notices are returned newest-first; tasks are returned soonest-due-first.

get_matter_documentsread onlysource verified78/100

Fetch every document attached to a matter (court notices, contracts, email attachments, connector imports). Returns kind, source connector, mime, bytes, and review status if available.

list_mattersread onlysource verified82/100

List matters in the current workspace. Defaults to open matters; pass status to filter (open | on_hold | closed).

list_upcoming_hearingsread onlysource verified80/100

List bankruptcy hearings (341 meetings, motion hearings, etc.) scheduled within the next N days, with case number, debtor, and trustee/judge if known.

search_matter_ragread onlyauthsource verified73/100

Semantic search over a matter's indexed document chunks (workspace + matter scoped). Returns the top-K chunks with similarity scores. Requires OPENAI_API_KEY on the server.

Issues (5)

HIGHresponse-shaper ↗

Output schemas not documented. Tools return JSON-stringified objects but field structure, types, and required fields are not declared. LLMs cannot reliably extract nested fields or plan downstream tool calls.

list_mattersget_matter_documentssearch_matter_raglist_upcoming_hearingsget_case_notice_timeline

MEDIUMtool-description ↗

Parameter descriptions incomplete or generic. 'limit' in list_matters and list_upcoming_hearings lack rationale (why 50 or 14 as defaults?). 'k' in search_matter_rag not described. LLMs cannot reason about appropriate values.

list_matterslist_upcoming_hearingssearch_matter_rag

HIGHrecovery-guide ↗

No error recovery guidance. Tools return bare JSON (e.g., {found: false}) without actionable next steps. If search_matter_rag fails due to missing OPENAI_API_KEY, the LLM receives no hint to check server config.

get_matter_documentssearch_matter_rag

Recommendations

Document output schemas for all tools. Declare field names, types (string, integer, array, object), and which fields are always present vs. optional. Example: 'Returns {matters: [{id: string, name: string, status: enum, createdAt: ISO8601}], count: integer, workspaceId: string}'.
Expand parameter descriptions to include rationale and constraints. E.g., 'limit: Maximum matters to return (1 - 200, default 50). Higher limits increase response time; use pagination for large workspaces.' Add similar guidance to 'k' in search_matter_rag.
Add error recovery guidance. When get_matter_documents returns {found: false}, append: 'Matter not found. Call list_matters to discover valid IDs.' When search_matter_rag fails, return: 'RAG search unavailable. Ensure OPENAI_API_KEY is set on the server. Fall back to get_matter_documents for document listing.'
Implement pagination for list_matters and list_upcoming_hearings. Add optional 'offset' or 'cursor' parameter and return 'next_cursor' or 'has_more' flag. This prevents context window exhaustion on large workspaces.
Add tool annotations (readOnlyHint) to all tools. Mark each as read-only so clients can optimize caching and avoid unnecessary re-execution. Example: {readOnlyHint: true} in tool metadata.

Score history

Overall score trend

First recorded score · v2 rubric

57/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	D	57	2026-07-28+	v2