Standard-library stdio MCP server for OpenCROW lifecycle operations
The opencrow-lifecycle-mcp server defines 7 tools for managing CTF challenge lifecycle workflows. All tools have explicit schemas with proper JSON Structure and descriptions present. However, there are significant gaps: parameter descriptions are inconsistent and often missing, output schemas are not documented, error handling is minimal, and tool names do not follow verb_noun patterns clearly (e.g., 'workflow_status' vs 'get_workflow_status'). Naming is functional but could be clearer for LLM disambiguation. The tools compose well for a document-centric workflow but lack the parameter granularity and output clarity expected of production-grade tools.
Append a sourced clarification without changing the Original Challenge.
Read bounded context from canonical lifecycle documents.
Append a reproducible attempt to CHANGELOG.md.
Append a stable, evidence-backed finding to FINDINGS.md.
Read the current challenge phase, lifecycle document state, and completion blockers.
Append a reproducible reconnaissance or unsolved-solve checkpoint to HANDOFF.md.
Append a verified solution or later verification/revision to WRITEUP.md.
Tool names do not consistently follow verb_noun pattern. Names like 'workflow_status', 'workflow_read', 'workflow_record_attempt' mix status-check (noun-only) with action verbs. Recommend 'get_workflow_status', 'read_workflow_documents', 'record_attempt', etc. to improve LLM verb-based tool selection.
Output schemas are not documented. Users and agents cannot know what fields each tool returns, which breaks multi-step tool chaining and forces agents to infer structure. Document return type as structured objects with field descriptions, especially for tools returning findings, attempts, and handoff data.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Parameter descriptions are sparse and non-standard. For example, 'workspace' parameter appears in all tools with no description of what it means or how to obtain it. The 'documents' enum in workflow_read is empty (items: {enum: []}), making the parameter unusable. Add descriptions for every parameter explaining its purpose, format, and constraints.
workflow_read has a 'documents' parameter with an empty enum (items: {enum: []}), making it impossible for agents to know what documents are valid. Either populate the enum with actual document types (e.g., 'README', 'CHANGELOG', 'FINDINGS', 'HANDOFF', 'WRITEUP') or remove the enum constraint and describe valid values in the parameter description.
No error guidance provided. Tools document neither error categories (retryable, user-fixable, fatal) nor recovery paths. For write tools (record_attempt, record_finding, etc.), agents need to know if a failure is transient or permanent, and what to do next.
Tool descriptions conflate implementation details (file names like CHANGELOG.md, FINDINGS.md) with user intent. Descriptions read like 'Append to CHANGELOG.md' rather than 'Record a completed or failed troubleshooting attempt for agent reference.' Rewrite to focus on WHAT agents should use the tool for and WHEN, not WHERE the data is stored.
Parameter 'status' in workflow_record_finding uses enum values 'confirmed|refuted|superseded', but no description explains when to use each. Add context: 'confirmed = evidence validates the finding; refuted = evidence disproves it; superseded = a newer finding replaces this one.'
workflow_record_finding requires a 'finding_id' that must match pattern ^F-[0-9]{4,}$ but provides no guidance on how agents should generate or obtain this ID. Should the tool auto-generate it, or must agents know the sequence? Document the expectation.