Loading ToolBench
Codebone provides 9 well-structured tools with complete JSON schemas and descriptions. All tools have clear names starting with 'codebone_' prefix and descriptions (avg ~120 chars). Input schemas are present and typed for all tools. However, descriptions lack LLM-optimized guidance on WHEN to use each tool vs alternatives, and error handling is minimal. Output schemas are not documented. Parameter descriptions are present but generic (e.g., 'token budget' without range/constraint guidance). No tool annotations (readOnlyHint, destructiveHint) despite clear risk profiles. Batch tool (codebone_batch) shows good composition pattern but lacks per-operation error handling.
Execute multiple codebone operations in a single batch request
Build symbol-aware context pack for a goal, optimized for token budget
Diagnostic tool to check codebone installation and configuration
Analyze impact of changes to a file or symbol on the rest of the codebase
Build or check status of the symbol index for fast lookups
Project map tool that generates a compact overview of directory structure and file organization
Read and extract code content from a file, optionally filtered by symbol or line range
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk profiles. codebone_index is marked WRITE but lacks destructiveHint; all READ_ONLY tools lack readOnlyHint. Agents cannot distinguish safe vs risky operations.
Output schemas not documented. Tools return complex nested structures (e.g., codebone_skeleton returns symbol trees, codebone_context returns file lists with metadata) but LLMs have no schema to plan downstream calls or extract fields.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2025-06-18+ | v2 |
Extract code skeleton showing structure, symbols, and signatures for a file or directory
Find and locate symbols by name, kind, or pattern within a codebase
Descriptions lack WHEN-to-use guidance. E.g., codebone_skeleton vs codebone_map both analyze structure; codebone_context vs codebone_skeleton both extract code. Descriptions do not explain when to prefer one over the other, forcing LLM to guess.
Parameter constraints not documented in descriptions. 'budget' appears in 5 tools but no guidance on valid range, units, or impact. 'limit' and 'offset' lack min/max. 'timeoutMs' lacks guidance on reasonable values.
Error handling minimal. No recovery guidance in tool descriptions. codebone_batch lacks per-operation error reporting, if 1 of 10 operations fails, unclear which one or why.