Mcp
auxiliar-mcp
MCP server that gives agents in-loop access to the auxiliar.ai Capability graph — task rankings, service intelligence, eval data, install commands, alternatives.
Verified: 2026-04-25 (vendor-self-reported)
When to use auxiliar-mcp
Choose if
Your agent needs in-loop access to ranked task templates (PDF OCR, NFS-e extraction, bookkeeping, etc.) and service intelligence (pricing, risks, compatibility).
Avoid if
You only need static tool recommendations and can hard-code them into the agent prompt — the MCP overhead isn't worth it.
Risk Flags
- LOW deprecation Tools 6–21 of the wrapped CLI agent are pending replacement with Capability-oriented tools (jobs/community/dm/profile-browse). MCP-server-level tools are stable.
- LOW vendor_lockin Server queries are answered against the auxiliar.ai-curated Capability index; ranking judgments reflect auxiliar.ai's editorial methodology rather than consensus across registries.
Cost
Type: Free · Free tier: Unlimited (npm package, runs locally; remote API calls hit api.auxiliar.ai which is rate-limited per IP)
Install
Default
claude mcp add auxiliar npx auxiliar-mcp
# Once installed, the agent can call:
solve_task(task_slug="nfs-e-extraction")
recommend_service(category="database")
list_solve_tasks()
Estimated time to first success: ~2 min
Claude desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"auxiliar": {
"command": "npx",
"args": ["-y", "auxiliar-mcp"]
}
}
}
Cursor
npx auxiliar-mcp
Add to .cursor/mcp.json or via Cursor's MCP settings UI
Dependencies
Minimum runtime: Node 18+
Composes with: auxiliar-solve, nfs-e-parser
Distribution
- npm
auxiliar-mcp- MCP Registry
io.github.Tlalvarez/auxiliar-mcp- Repository
- https://github.com/Tlalvarez/Auxiliar-ai
- License
- MIT