Siyuan Agent

Interact with SiYuan notes via direct HTTP API. Use when reading, writing, searching, or managing SiYuan blocks, documents, notebooks, attributes, assets, or...

This skill appears to implement exactly what it claims (a SiYuan HTTP API CLI) and enforces reasonable write and SQL restrictions, but the package metadata omitted the required SIYUAN_TOKEN. Before installing: (1) Confirm you set SIYUAN_TOKEN — the tool will exit without it; (2) Keep SIYUAN_BASE at the default localhost unless you intentionally want to point it elsewhere (setting it to a remote URL will send your token and data to that host); (3) Inspect the small JS files (they're included and readable) to verify no unexpected network targets; (4) Prefer minimizing agent autonomy or restricting the skill's use until you trust the publisher; and (5) Verify the publisher/slug/source — the registry metadata inconsistency could be a packaging error, but treat it as a warning.

Type: OpenClaw Skill Name: siyuan-agent Version: 2.0.6 The siyuan-agent skill is a legitimate tool for interacting with the SiYuan note-taking application via its HTTP API. It implements several safety features, including a mandatory 'write=true' flag for destructive operations, a blocklist of sensitive notebook management endpoints (FORBIDDEN_PATHS in siyuan.js), and basic SQL validation to restrict queries to SELECT statements. The code is transparent, uses native fetch without external dependencies, and only communicates with the user-configured SIYUAN_BASE endpoint.