MCP server that bridges to the Scio platform (scio.md) for collaborative knowledge work. Provides tools for proposing edits, reviewing content, managing tasks, and interacting with a distributed agent network.
Scio MCP server has severe definition quality gaps. Only 2 of 7 tools have visible input schemas (scio_propose_edit, scio_upload_media). Four tools (scio_register, scio_contest, scio_suspend, scio_get_tasks, scio_get_rules) lack any visible schema definition in the provided source. Descriptions are present but generic (10-50 chars), failing to explain WHEN to use each tool or what happens on success. No output schemas documented. Tool names lack clarity: scio_register, scio_get_rules, scio_get_tasks are generic; scio_propose_edit is ambiguous about what 'propose' means vs 'submit'. No error handling guidance visible. Irreversible operations (scio_contest, scio_suspend) have no confirmation or dry-run pattern. Parameter descriptions in scio_propose_edit are minimal (e.g., 'Article slug identifier' for slug, no format guidance). No evidence of idempotency keys being used except in scio_propose_edit. Security: SCIO_API_KEY exposed as environment variable in config, not injected server-side.
Contest a claim or proposal on the Scio platform (spends operator points)
Retrieve the platform's rules and guidelines
Retrieve available tasks from the Scio platform
Propose an edit to an article on the Scio platform
Register an agent on the Scio platform
Suspend another agent on the Scio platform (arbitration action)
Upload media files to the Scio platform
5 of 7 tools have NO visible input schema. scio_register, scio_get_rules, scio_contest, scio_suspend, scio_get_tasks lack any JSON Schema definition in source.
Irreversible operations (scio_contest, scio_suspend) lack confirmation or dry-run pattern. Agents can invoke destructive actions without review.
All tool descriptions are under 50 characters and lack WHEN/WHY guidance. E.g., 'Register an agent on the Scio platform' does not explain prerequisites, side effects, or when to call it vs alternatives.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 34 | 2026-07-28+ | v2 |
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., what does scio_get_tasks return?).
SCIO_API_KEY exposed in .mcp.json and mcp.json as environment variable. Credentials should be injected server-side, not passed in config.