Complete MCP server with agentic tools, artifact management, ARC gap detection, and semantic search capabilities. Provides tools for code search, file operations, artifact approval workflows, and conceptual gap analysis.
VETKA MCP exhibits significant definition quality gaps. While 16 tools are registered with descriptions and input schemas, most descriptions are generic or lack actionable guidance. Parameter descriptions are sparse or missing entirely. No output schemas are documented. Error handling is absent, tools provide no recovery guidance. Tool naming is inconsistent (vetka_* prefix on some, bare verbs on others). Critical security tools like execute_code lack proper constraints and confirmation patterns. The semantic_search, knowledge_graph, and arc_gap tools reference external systems (Qdrant, knowledge bases) without documenting expected response structures or failure modes. Artifact management tools (approve/reject/edit) lack idempotency markers and confirmation workflows. Overall, definitions read as API pass-throughs rather than LLM-optimized tool interfaces.
Execute shell commands safely with optional sandbox. Phase 20: Enhanced with macOS sandbox-exec support. Security features: Case-insensitive pattern matching, Blocks absolute paths to dangerous binaries, Blocks shell metacharacters for command injection
Check VETKA server health and status. Returns server uptime, memory usage, and component status.
Query VETKA knowledge graph. Returns relationships and connections between concepts, files, and entities.
List files in a directory with optional filtering. Returns file metadata including size, modification time, and type.
Get VETKA metrics and statistics. Returns code metrics, performance data, and system statistics.
Read file content from VETKA project. Returns full file content with line numbers. Supports any text file within the VETKA project directory.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. semantic_search, knowledge_graph, metrics, health_check lack any response structure definition.
execute_code tool (IRREVERSIBLE risk) lacks confirmation/dry-run pattern and timeout validation. No guidance on sandbox escape risks or command injection prevention despite mentioning 'security features'. LLM could execute destructive commands without safeguards.
Artifact management tools (vetka_edit_artifact, vetka_approve_artifact, vetka_reject_artifact) lack idempotency markers and error recovery guidance. No documentation of what happens if approval is called twice or if artifact_id is invalid.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Search for text/pattern in codebase using grep. Returns matching lines with file paths.
Search for files by name or pattern. Returns matching file paths and metadata.
Semantic search in VETKA knowledge base using Qdrant. Searches for concepts, ideas, or topics across all indexed documents using vector similarity search.
Get VETKA 3D tree structure. Returns the hierarchical file/folder structure used by VETKA's 3D visualization. Supports summary or full tree output.
Approve pending artifact and save to disk
Extract key concepts from text using pattern matching (fast, no LLM calls)
Analyze prompt/context for conceptual gaps using ARC methodology. Returns suggestions for missing connections, patterns, or concepts that could improve the quality of agent responses.
Edit artifact content and submit for approval
List artifacts by status
Reject artifact with feedback
Parameter descriptions are sparse or missing. 'metric_type' in metrics tool lists valid values ('code, performance, system') in description but not as enum constraint. 'depth' in knowledge_graph lacks range bounds. 'timeout' in execute_code lacks validation rules.
No error handling or recovery guidance in any tool description. Tools provide no actionable error messages (e.g., 'File not found. Try search_files() first' or 'Invalid metric_type, must be code|performance|system'). LLMs cannot self-correct on failures.