Bridge WeChat messages into Claude Code sessions via the official ilink API. Enables WeChat users to interact with Claude Code through a WeChat bot interface.
Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and error handling. Both tools expose context_token as a parameter (potential security concern), lack output schema documentation, and descriptions are minimal (under 100 chars). No error recovery guidance, no idempotency hints, and no validation constraints documented. Naming is clear (send_text, send_image) but parameter descriptions are sparse. This server falls into the 'poor' category typical of community STDIO servers.
Send an image file to a WeChat user or group via CDN upload
Send a text message reply to a WeChat user or group
context_token exposed as tool parameter, security risk. Tokens should never appear in tool parameters; they are logged in traces and prompt history. Use server-side secret injection instead.
No output schema documented. LLMs cannot plan downstream calls or extract success/failure signals. Document what send_text and send_image return (e.g., message_id, timestamp, status).
Parameter descriptions are minimal (under 20 chars for context_token). LLMs cannot infer what context_token is, when to populate it, or how to obtain it. Expand descriptions with format, source, and usage guidance.
No error handling or recovery guidance. If send_text fails (invalid user_id, network error, rate limit), the LLM has no guidance on retry strategy or next steps. Add error classification and actionable recovery hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | 2026-07-28+ | v2 |
No idempotency or confirmation pattern. Sending a message is irreversible, if an LLM retries on ambiguous failure, duplicate messages result. Add idempotent hints or a dry-run/confirmation step.