Production AI coding agent for your terminal - multi-provider support, context management, MCP integration
Two tools with minimal documentation. edit_file and write_file have basic descriptions (50-60 chars, within acceptable range) but lack critical parameter-level documentation. Input schemas are visible but parameters lack descriptions, a hard violation of the rubric. No output schema documented. No error handling guidance. No tool annotations. Naming is clear (verb_noun pattern), but the implementation is far below production baseline.
Make a surgical edit by replacing a unique string in a file (with LSP diagnostics)
Write content to a file (creates or overwrites) with LSP diagnostics
All parameters lack descriptions. 'path', 'old_string', 'new_string', and 'content' have no explanatory text. LLMs cannot infer parameter meaning from names alone.
No output schema documented for either tool. Callers do not know what fields to expect in the response, breaking downstream tool chaining.
No error handling guidance. Tools that modify files (WRITE risk) must document failure modes and recovery steps. E.g., 'If old_string not found, returns error with line count and preview.'
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Both tools are marked WRITE risk but lack machine-readable hints for agent planning.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 52 | 2026-07-28+ | v2 |
edit_file requires 'old_string' to be unique in the file. This constraint is not documented in the parameter description, forcing LLMs to guess or fail silently.