Agent-native · MCP + API

Drive FounderReply from Claude, Cursor, or your own code.

FounderReply ships a real MCP server and a code-level ToS-compliance controller. Every programmatic call runs through the same hard caps as the app — the agent surface can never exceed what a platform's policy allows.

MCP endpoint Application API

The agent surface

The MCP endpoint

MCP (Model Context Protocol) lets an AI agent enumerate and call tools over a standard wire format. FounderReply exposes its agent at a single endpoint — POST /api/mcp — speaking JSON-RPC 2.0.

create_personaGenerate and persist a brand persona / content strategy from a company-focus brief.

get_personaRead the workspace’s latest persona, or null if none exists yet.

generate_content_planBuild a rolling content calendar of planned posts over the next horizonDays.

list_scheduled_postsList scheduled posts, optionally filtered by status.

set_autonomySet the requested autonomy mode for an action — the ToS hard cap still applies at decision time.

get_analyticsEngagement overview over a trailing window, with a per-content-pillar rollup.

find_growth_opportunitiesDiscover public conversations the brand could add value to, and persist them.

list_opportunitiesList engagement opportunities, optionally filtered by status.

tools/list — enumerate the tools

POST /api/mcp
Authorization: Bearer fr_live_…

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list"
}

tools/call — invoke a tool

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "set_autonomy",
    "arguments": {
      "action": "reply",
      "mode": "approve"
    }
  }
}

Scoped and capped by design.

A key is bound to one workspace — tools never take a workspace id, so a key can only ever operate its own. And set_autonomy records what you request; the platform-ToS hard cap is still applied at decision time, so a key can never push an action past what a platform allows.

Get an API key

Scope & roadmap: today the endpoint supports the two core MCP methods — tools/list and tools/call — over a single HTTP request/response. The full MCP SSE / streamable-HTTP transport (server-push, notifications, sessions) is on the roadmap. fr_live_ keys authenticate this MCP endpoint.

The application surface

The v1 API

The /api/v1/** routes power the FounderReply app itself. Today they authenticate with your logged-in session, not an API key — for headless automation, drive the MCP endpoint above. The surface covers:

/publishPush an approved draft to a connected account via its official API.

/approvalsThe human-in-the-loop queue — approve, edit, or skip drafted replies.

/accountsConnect / disconnect your own platform accounts over OAuth.

/personasRead and manage the brand persona / content strategy.

/scheduled-postsInspect and manage the planned content calendar.

/analyticsEngagement metrics rolled up by content pillar.

/autonomyRead / set the per-action autonomy dial (still ToS-capped).

/growthOpportunity discovery and the act-on-opportunity flow.

Under the hood

Five official-API connectors

Each connector talks to a platform's official API and declares its real capabilities — we never fabricate a capability a platform doesn't offer.

RedditPublish · reply to comments · comment on others’ threads · DM · mentions · metrics (score, comments).

XPublish · reply · comment on others · DM · mentions · public metrics · audience. BYO-credential, Pro-tier.

InstagramPublish · reply on your own posts · DM · webhooks · metrics · audience. No API for commenting on others.

FacebookPublish · reply on your own posts · DM · webhooks · Page metrics · audience. No API for commenting on others.

LinkedInPublish · reply on your own posts · social metrics. No third-party DM or comment-on-others API.

Every action a connector takes — own-account or otherwise — is gated by the autonomy controller first, so commenting on others' threads is capped (and on Reddit, off by default) regardless of how a call arrives.