Multi-agent coordination bridge between Claude desktop and Claude Code via file-backed state and real-time browser observability.
Two tools with minimal but present descriptions and empty input schemas. check_silas_messages has a detailed description (180 chars) explaining context and behavior; send_to_silas similarly well-described (165 chars). However, both tools lack input parameter descriptions. check_silas_messages accepts no parameters (empty object), which is acceptable. send_to_silas accepts a 'content' parameter with a type but no description visible in the schema definition shown. No output schemas documented. No error handling guidance. No tool annotations (readOnlyHint/destructiveHint). Naming is clear and verb-based (check_, send_), following pattern:tool conventions. Overall structure is minimal but functional for a simple coordination bridge.
Read the latest message the Claude Code instance has posted via Claude Sync. Returns the message content and server-side timestamp. If no message is waiting, returns a clear empty-inbox signal. 'Silas' is the naming convention for the Claude Code side in this project (see README) — this tool is how you receive context from that instance.
Post a message to the Claude Code instance via Claude Sync. The target side reads from the same state file and will see the message on its next tool call or GET. Use for handoffs, scope approvals, and context sharing. 'Silas' is the project-local naming for the Claude Code side (see README).
No output schemas documented. LLMs cannot plan downstream operations or extract structured data without knowing what fields to expect.
send_to_silas parameter 'content' lacks a description. LLMs cannot infer constraints, format, or length limits from the name alone.
No error handling guidance. If a message fails to post or read, LLMs have no recovery path. Errors should indicate whether to retry, ask the user, or abort.
No tool annotations (readOnlyHint, destructiveHint). check_silas_messages is read-only; send_to_silas is destructive (modifies state). Annotations help LLMs reason about side effects and retry safety.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
No pagination or result limits documented. If message history grows large, responses could bloat context. check_silas_messages should document max message size and history retention policy.