MCP server for interacting with Gitopia blockchain-based Git repository platform. Provides tools for repository management, issue tracking, pull requests, bounties, DAO governance, and batch transactions.
Server has 11 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, mostly under the 50-200 char LLM-optimized range), and input schemas are present with type definitions. However, output schemas are completely undocumented, LLMs cannot plan downstream calls or extract return values. Parameter descriptions are minimal (e.g., 'The ID of the pending transaction to confirm' for confirm_transaction). No error handling guidance, no recovery paths, and no indication of which operations are idempotent or retryable. Tool naming is clear and verb-based (confirm_, reject_, list_, create_, update_, close_, delete_, get_), but composition issues exist: batch_execute conflates multiple concerns into one tool, and no tool chains are documented (e.g., create_bounty returns what ID? How does it chain to update_bounty?). Security: no evidence of permission gates, audit logging, or secret injection patterns. Overall, definitions are skeletal, sufficient to invoke but insufficient for reliable agent planning.
Executes multiple operations in a single on-chain transaction
Closes a bounty
Broadcasts a previously held pending transaction
Creates a bounty for an issue
Deletes a bounty
Retrieves a specific bounty
Lists bounties
Lists commits for a repository branch
No output schemas documented. LLMs cannot infer return types, field names, or structure. Agents cannot plan downstream tool calls or extract required IDs for chaining (e.g., create_bounty returns bounty_id, but this is not documented).
Descriptions are too brief (10-50 chars) and lack context. Missing WHEN to use each tool, prerequisites, and side effects. E.g., 'Broadcasts a previously held pending transaction' does not explain what 'held' means, when to call this vs reject_transaction, or what happens on success.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | <=2025-11-25 | v2 |
Lists all non-expired pending transactions for the current session
Cancels a previously held pending transaction
Updates a bounty expiry
No error handling guidance. Tools do not indicate which errors are retryable, which require user intervention, or how to recover. E.g., if confirm_transaction fails because the transaction expired, the LLM has no guidance on what to do next.
batch_execute conflates multiple concerns (execute N operations in one call). This violates single-responsibility principle. Agents cannot compose individual operations; they must pre-plan the entire batch. Consider splitting into separate tools or documenting strict composition rules.
Destructive operations (delete_bounty) lack confirmation or dry-run support. No evidence of permission gates or audit logging. Agents can delete bounties without safeguards.