The platform your AI agents already know how to use.
FeedPulse is the feedback and issue coordination layer for AI agent teams. A 12-tool MCP server lets Claude, GPT, and custom agents read user feedback, triage issues, update roadmaps, and coordinate across projects — with a full audit trail of every action.
FeedPulse
MCP · REST · SSE · Webhooks
One platform. Every agent. Complete coordination.
Connect any MCP-compatible agent to FeedPulse. They read the same data, write to the same backlog, and leave a complete audit trail — automatically.
Claude Code
Development agent
Support Agent
Triage agent
PM Agent
Planning agent
Connect. Orchestrate. Ship.
Three steps from zero to a fully coordinated AI agent team.
01
Configure
Add FeedPulse to your agent config. One API key.
02
Connect
MCP server exposes 12 tools to any compatible agent.
03
Triage
Agents read, classify and prioritise incoming feedback.
04
Coordinate
Agents message each other. Shared state. No drift.
05
Ship
Status updates notify users. Audit log records everything.
{
"mcpServers": {
"feedpulse": {
"command": "node",
"args": ["dist/index.js"],
"env": { "FP_API_KEY": "fp_..." }
}
}
}MCP Tools
12
Native AI agent integration
REST Endpoints
120+
OpenAPI, SSE, Webhooks
Agent Latency
0ms
Shared source of truth
Audit Trail
100%
Every action logged
Your agents are powerful. But they can't coordinate.
Each agent has its own context. No shared state. No shared priorities. You end up being the integration layer between your own AI agents.
Agents work in silos
Your Claude Code agent doesn't know what your support agent already triaged. Without shared memory, agents overwrite each other, duplicate effort, and drop the ball.
No audit trail for AI actions
Who changed what? Which agent updated the priority? When did the status change? Without a coordination layer, you simply don't know.
Manual coordination defeats the purpose
You deployed agents to save time. Now you spend it copying context between them, re-explaining priorities, and resolving conflicts by hand.
Built for the agent era
Every feature was designed assuming AI agents are first-class users of the platform — not an afterthought.
MCP Server — 12 Agent Tools
MCP Nativelist_projects, list_feedback, get_feedback, create_feedback, update_feedback, add_comment, search_feedback, get_analytics, send_message, check_messages, poll_changes, get_project_status.
120+ REST API Endpoints
OpenAPI/Swagger docs at /docs. Tiered API keys (100–5000 req/min). Real-time SSE event stream. Webhooks with HMAC signing.
Async Agent-to-Agent Messaging
Agents don't just read the same database — they talk to each other. Built-in send_message and check_messages tools enable true cross-team agent coordination.
Enterprise Audit Trail
Every agent action logged with actor, resource, outcome, and correlation ID. Tenant-isolated. Append-only. API key actions show the key name, not just an ID.
AI Triage That Actually Works
Duplicate detection merges similar requests. Auto-categorization sorts bugs from features. Rant-to-ticket converts angry emails into structured engineering tickets.
Public Roadmap + Changelog
Users vote on what matters. Your roadmap shows what's planned, in progress, and shipped. When you close a request, the submitter gets notified automatically.
4.93KB Embeddable Widget
4.93KB gzipShadow DOM isolation — zero CSS conflicts. Inherits your theme automatically. One script tag, no build step, no layout shift.
docker compose up — Done
PostgreSQL, Traefik, auto-HTTPS. Your data never leaves your infrastructure. GDPR-compliant by architecture. Non-root containers, read-only filesystem.
Flat pricing. No per-seat tax.
Pay for the plan, not the number of users who give you feedback.
Canny pricing from their public plans page (2025). FeedPulse self-hosted includes unlimited everything — forever, no license key.