Onicode provides 6 tools with complete input schemas and descriptions. Naming follows verb_noun convention (ctx_execute, ctx_search, ctx_index, ctx_batch, ctx_stats, ctx_fetch). Descriptions are substantive (100-250 chars), explaining WHAT each tool does and WHEN to use it. However, output schemas are not documented, LLMs cannot predict return structure. Error handling is absent; no guidance on retryability or recovery. Tool composition is sound (each does one thing), but missing pagination parameters on ctx_search and ctx_batch despite returning multiple results. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk profiles (ctx_execute is IRREVERSIBLE, ctx_index is WRITE). Parameter descriptions are present but lack format constraints (e.g., timeout range, limit bounds). Overall: solid definitions with gaps in output documentation and error guidance.

• Document output schemas for all tools. Example for ctx_search: 'Returns {results: [{content: string, source: string, score: number}], total: number, nextCursor?: string}'.
• Add error handling guidance to tool descriptions. Example: 'If timeout, increase timeout parameter and retry. If index fails, verify content is valid UTF-8 text or JSON.'
• Annotate tools with risk hints: ctx_execute → destructiveHint=true (code execution), ctx_index → destructiveHint=false (write-only), ctx_fetch → readOnlyHint=true.
• Add parameter constraints: timeout (min=1000, max=300000), limit (min=1, max=100), language enum validation in ctx_execute.
• Add pagination to ctx_search and ctx_batch: include offset/limit per query, return total count and nextCursor for continuation.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry