A compiler and CLI tool for building AI agents with support for multiple execution blocks (webhook, local, mcp, builtin, client, provider_hosted, knowledge, slng) and code generation for LiveKit, Pipecat, and SLNG platforms
Both tools have minimal descriptions (under 100 chars) and lack parameter-level documentation. Schemas are present but sparse. Tool names lack action verbs (book_appointment, cancel_appointment are acceptable but generic). No error handling guidance, no output schema documentation, and no indication of idempotency or side effects. This is a test/mock server with placeholder tool definitions.
Mock tool that books an appointment for a customer
Mock tool that cancels an appointment
Parameter descriptions missing. Input schema defines customer_id, service, slot_id but provides no guidance on format, constraints, or valid values. LLMs cannot infer whether service is an enum or free-form string.
No output schema documented. Callers do not know what fields are returned (e.g., confirmation_id, timestamp, status). This breaks tool chaining and forces LLMs to guess downstream parameter names.
Tool descriptions do not indicate side effects or idempotency. 'Mock tool that books an appointment' does not tell the LLM whether repeated calls create duplicates or are safe to retry.
No error handling guidance. No documentation of what happens if customer_id is invalid, slot_id is unavailable, or appointment_id does not exist. LLMs have no recovery path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | 2026-07-28+ | v2 |
Tool names are generic. 'book_appointment' and 'cancel_appointment' lack context about what system they operate on. If multiple appointment systems exist, names should disambiguate (e.g., calendar_book_appointment).