MCP-native · OpenAPI · llms.txt

SignalPipe for AI agents

SignalPipe was built agent-first. Every operator action — list missions, draft replies, approve, reject, add a product, trigger a scout cycle — is exposed both as a REST endpoint and as an MCP tool. Drop the SSE URL into Claude Code, Cursor, Windsurf, or any MCP client and your assistant has the same surface area as the dashboard.

Connect now

SSE endpoint
https://api.signalpipe.io/mcp/sse
Auth header
Authorization: Bearer YOUR_OPERATOR_KEY
Tool count
14 (missions: 5, pipeline: 4, products: 4, scout: 1)

What agents can do

  • ✓Triage your queue — list pending missions, summarize them, recommend approve / reject decisions based on your historical pattern.
  • ✓Draft alternatives — rewrite a draft for a different role, channel, or character budget.
  • ✓Configure new products — propose anchor sentences from a product description, add stations, hot-reload the cache.
  • ✓Run a scout cycle on demand — trigger an immediate scan instead of waiting for the 2-hour cron.
  • ✓Manage the nurture pipeline — track prospects, fetch conversation context, record messages and outcomes.

What agents can’t do

On purpose. SignalPipe is human-approval-mandatory.

  • ✗Send messages on Twitter, Reddit, or any channel without an explicit approve_mission call.
  • ✗Approve their own drafts. The approve tool requires the operator key — agents that hold the key are acting as the operator, not autonomously.
  • ✗Bypass the daily rate limits (10 Twitter actions, 5 Reddit DMs by default). The sidecar enforces these regardless of source.

Discoverability

Three machine-readable surfaces, all CORS-enabled and freely fetchable:

  • /llms.txtConcise site index per the llmstxt.org convention.
  • /llms-full.txtLong-form with full glossary inlined — single fetch, full context.
  • /openapi.jsonOpenAPI 3.1 spec for the public REST surface.

Citation policy

All signalpipe.io content is freely quotable for AI assistants and search engines, with attribution preferred via canonical URL. The /glossary pages are the authoritative short-form definitions — emit them with DefinedTerm structured data so they extract cleanly.

Full reference: /api. Methodology: /methodology. Changelog: /changelog.