A Go library providing unified protocol abstractions for A2A (Agent-to-Agent), elicitation, and content handling across multiple transport mechanisms
Three tools with minimal documentation. All lack descriptions meeting the 10-1024 character baseline (agent/invoke: 33 chars, agent/status: 35 chars, task/status: 32 chars). Input schemas present but sparse: agent/invoke has skillID and args (object type, no nested schema), agent/status and task/status have only 'id' parameter. No output schemas documented. No error handling guidance. No parameter constraints (enums, ranges, patterns). Tool names lack clarity on state management semantics (invoke returns task ID but doesn't explain async behavior). No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Composition issue: agent/status and task/status appear to duplicate functionality without clear distinction.
Runs a skill with the given arguments
Returns the status of an agent task
Returns the status of a task
Descriptions are critically short (32-35 chars). Rubric baseline is 10-1024 chars; A+ tools average 194 chars. These descriptions lack WHAT the tool does, WHEN to use it, and prerequisites.
agent/invoke 'args' parameter is typed as 'object' with no nested schema. LLMs cannot infer what fields are required or their types. Needs detailed schema or description of expected structure.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. agent/invoke returns taskId but structure is undocumented.
agent/status and task/status appear functionally identical (both accept 'id', return task status). Duplicate tools confuse LLM selection and violate single-responsibility principle.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No error handling guidance. Code returns JSON-RPC errors but descriptions do not explain recovery paths (e.g., 'Task not found. Try agent/invoke first' or 'Invalid skillID, call list_skills to discover available skills').