Better web and X/Twitter search for Codex and Claude Code, powered by Grok CLI.
Two tools with complete schemas and reasonable descriptions. Both tools have proper input validation via Zod and error handling. However, descriptions are generic (under 100 chars) and lack actionable context for LLM selection. Parameter descriptions are present but minimal. No output schema documentation. Tools lack idempotency hints and recovery guidance in error messages. Naming is clear (verb_noun pattern) but descriptions don't explain WHEN to use each tool vs. the other, or what distinguishes them.
Search the web through an isolated Grok CLI profile. Does not use xAI API credits.
Search X/Twitter through an isolated Grok CLI profile. Does not use xAI API credits.
Output schema not documented. Tools return JSON-stringified text without specifying field structure (e.g., result count, result format, pagination). LLMs cannot plan downstream operations or extract specific fields.
Tool descriptions lack differentiation and context. Both descriptions are ~80 chars and do not explain WHEN to use X search vs. web search, or what 'isolated Grok CLI profile' means to an LLM. Descriptions should be 50-200 chars with clear selection criteria.
Error messages are generic. errorResult() returns raw error text without recovery guidance. E.g., 'Invalid input' does not tell the LLM what to fix or retry. Should include constraint violations and next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, idempotentHint). Both tools are read-only and idempotent, but this is not declared in the schema. LLMs cannot infer safety properties and may avoid retrying on transient failures.
Parameter descriptions are minimal (10-30 chars). E.g., 'Optional X handles to prioritize, e.g. @grok.' lacks format guidance. Should specify: 'Comma-separated X handles (e.g., @grok, @elonmusk) to boost results from these accounts.'