LLM pricing for coding agents — every price carries its source and the date it was confirmed.
Three tools with complete input schemas and descriptions. Naming follows verb_noun pattern (get_, estimate_, find_). Descriptions are adequate (100-200 chars) but lack actionable context for LLM selection. Parameters have types and descriptions, but lack constraints (enums, ranges, defaults). Output schemas are not documented. Error handling is minimal. No tool annotations present.
Estimate the cost of running a conversation with specified models, accounting for conversation structure, caching, reasoning multipliers, and batch API usage
Find cheaper models that meet a minimum capability threshold compared to a reference model
Resolve a user-supplied model name to a catalog model and return pricing information with provenance
Output schemas not documented. LLMs cannot plan downstream operations or extract required fields without knowing response structure.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All three tools are read-only but this is not declared in the schema.
Numeric parameters lack constraints (min/max). 'system_tokens', 'user_tokens', 'output_tokens', 'turns', 'conversations_per_day' are unbounded, allowing LLMs to pass absurd values (negative, zero, millions).
Descriptions lack actionable context. 'Resolve a user-supplied model name...' does not explain WHEN to call this vs estimate_cost, or what happens if the model is not found.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
No error handling guidance. If a model name is invalid or not found, the LLM has no recovery path documented.