Prodflow
K
TL
C
Claude Code· reading
JD
4 collaborators on this page

Write once.
Every agent knows.

Prodflow is where engineering teams write docs and skills that both humans and AI agents can read. One shared canvas.

ClaudeCursorAmpAmpCodexCodexCline

How it works

K
editing

Your team's knowledge lives here. AI agents read it, update it, and propose changes. Humans review and merge.

Shared knowledge base
Pages with hierarchy, real-time editing, and branch-based review. A wiki your AI tools already understand.
AI-proposed updates
Connect Slack and GitHub. When conversations outpace your docs, AI proposes changes as reviewable branches.
MCP context layer
Your coding agents query Prodflow for live context. Architecture, specs, decisions — always current.
Publishable skills
Create skill files that teach agents how your team works. Install across any agent with one command.

Branch-based review

K
proposed changes

When your Slack conversations or GitHub PRs outpace the docs, AI proposes changes as branches. Review the diff, check the evidence, merge when ready — like a PR for your documentation.

main
add-rate-limiting
AI proposed · 2 min ago
## Authentication
v3 uses scoped API keys with OAuth 2.1 PKCE.
 
+ ## Rate Limiting
+ v3 introduces per-key rate limiting using a token
+ bucket algorithm. Default: 1,000 req/min (Pro),
+ 100 req/min (Free). Returns 429 with Retry-After.
 
- ## Migration Timeline
- Phase 1 target: March 2026
+ ## Migration Timeline
+ Phase 1 shipped. Phase 2 target: April 2026.
Evidence: 3 Slack messages, 1 PR

Skill registry

Y
you

Skills teach AI agents your team's workflows. Write on Prodflow, publish with one click, install everywhere.

Published Skill
npx @prodflow/cli skills add @prodflow/code-review
Public
Description
Review code for quality, security, and maintainability.

Version
1.2.0

Tools
Read · Grep · Glob · Bash
npx @prodflow/cli skills add @prodflow/code-review

Connect your agent

Add Prodflow as an MCP server. OAuth handles auth — paste the URL and go.

1

Add the MCP server

Paste the URL into your AI tool's MCP config.

2

Authenticate once

OAuth opens in your browser. Approve and you're connected.

3

Your agent is context-aware

Architecture, specs, skills — in every prompt.

mcp-config.json
// Claude Code, Cursor, or any MCP client
{
  "mcpServers": {
    "prodflow": {
      "type": "http",
      "url": "https://api.prodflow.co/mcp"
    }
  }
}

Your docs. Your skills.
Every agent, instantly.

Stop repeating context. Share knowledge that works for humans and AI alike.

Start for free →

Or stay in the loop