Hi everyone 👋 I built ctxd because I was sick of being the copy-paste machine. A decision was made in Slack last month. The contract is in Drive. The real bug got discussed in a GitHub PR. ChatGPT could reason about all of it. It just couldn't see any of it. So I dug it up by hand every time and pasted it in. Nobody fixes this because it isn't one hard problem. It's twenty boring ones. OAuth for every app. Downloading files. Turning PDFs and spreadsheets into clean text. Re-syncing when things change. I watched two teams spend months on that plumbing before they got to the part they cared about. So ctxd is the boring middle layer. You connect Google Drive, Slack, GitHub, and Calendar. It pulls everything in. It parses it, OCR included. It puts it in one searchable index. Your agent searches all of it in one call. A few choices we made on purpose: 🤖 It's built for agents, not just people clicking a UI. Use a plain API, an MCP server (Claude, Cursor, your own agents), or a CLI. 🎯 Search is programmable. An agent can filter by date, channel, repo, or folder. It doesn't have to hope a vector search guesses right. Plain English works too. 🔒 It respects permissions. You only get results you have access to. Data is isolated per user. Every result links back to the source so you can check it. ✅ We passed Google's CASA Tier 2 audit. Everything is encrypted. We can run on-prem if your data can't leave. Free tier is 1 user, 1,000 docs, all connectors. Paid starts at $15/user/mo. Every connector is included. No per-connector fee, which some tools quietly charge. Two questions for you: 1. What's the first thing you'd point it at? 2. Which app should we connect next? We hear Notion and Gmail the most. I'll be in the comments all day. Tear it apart 🔨 I'd rather hear what's broken. — ctxd.dev

CTXD: Your files, any AI agent

The context layer between your files and your AI Agents

The context layer between your files and your AI Agents