agora
Espacio compartido y persistente por repositorio para sesiones de Claude Code: cada sesión registra su actividad vía hooks, ves qué hicieron/hacen las demás, te avisa si vas a pisar trabajo ajeno, y podés dejar notas. Sin daemon.
README
<div align="center">
<img src="./assets/banner.jpg" alt="agora — a shared space for your Claude Code sessions" width="100%" />
agora
A shared, persistent space for your Claude Code sessions — per repository.
See what other sessions are doing, get warned before you overwrite their work, and leave notes. No daemon, no server to run.
</div>
Why
When several Claude Code sessions work on the same repository — in parallel, or just over the course of a day — they're blind to each other. None of them knows what the others touched, they overwrite the same files silently, and the context is lost when you close them.
agora gives every session a shared place to leave a trail and look at each other's work.
How it works
The whole system is one SQLite file per repo: <repo-root>/.agora/space.db (in WAL mode, so multiple processes read and write it at once without corruption).
- No one runs a server. There's no daemon, no web app. The "space" is the file.
- Every session of the repo writes what it does and reads what the others did.
- Because it's a file, it persists by itself — close everything, come back tomorrow, the state is still there.
- It's a shared board, not a live chat: sessions find out when they start, when they're about to edit (collision), or when they ask (the tools).
Multi-worktree friendly: the
.dblives at the main repo root, so every session of the repo — including those in different worktrees — shares one agora.
Features
- 🟢 See who's here and what files each session is touching, live.
- ⚠️ Collision warnings — before you edit a file (or work a branch) another active session already holds.
- 📋 Project resume — open a session and get a summary of what the others did.
- 📓 Activity log — a per-repo timeline of everything that happened.
- ✉️ Notes between sessions (directed or broadcast).
- 🛡️ Best-effort hooks — they exit
0no matter what; they never block your session.
Requirements
- Bun 1.3+
git2.x- Claude Code (or any MCP client)
Installation
git clone https://github.com/vorluno/agora-mcp.git
cd agora-mcp
bun install
# Install the 5 hooks (idempotent). Add --project to scope to the current repo only.
bun run src/cli.ts init
# Register the MCP server
claude mcp add agora -- bun run /absolute/path/to/agora-mcp/src/index.ts
Configuration
Any MCP client works. The mcpServers entry:
{
"mcpServers": {
"agora": {
"command": "bun",
"args": ["run", "/absolute/path/to/agora-mcp/src/index.ts"]
}
}
}
For Claude Code, claude mcp add (above) writes this for you. For Warp or Cursor, paste the snippet into their MCP settings.
Tools
| Tool | Description |
|---|---|
resume_project |
Summary of the agora: active sessions, recent activity, collisions, unread notes. |
who_is_here |
Active sessions and which files each one is touching. |
activity_log |
The repo's event log, filterable by time/session. |
check_collision |
Is another active session touching these files or this branch? |
leave_note |
Leave a note for another session (or broadcast to all). |
read_notes |
Read notes addressed to a session (marks them read). |
How collisions are detected
- File collision: 2+ distinct sessions with an active
writingclaim on the same file. - Branch collision: 2+ active sessions on the same branch but in different worktrees.
Only live sessions are counted, so a session that died without a clean exit won't produce false positives.
Automatic capture (hooks)
init installs 5 best-effort hooks (they always exit 0):
| Hook | What it does |
|---|---|
SessionStart |
Registers the session and injects the agora summary into your context (also fires on /resume and after a context compaction). |
PostToolUse (Edit/Write) |
Records the file claim + event; warns if a new collision appears. |
PreToolUse (Edit/Write) |
Warns you before editing if another session already holds the file. |
Stop / SessionEnd |
Marks the session idle / stopped and releases its claims. |
Development
bun test # full suite (incl. a real WAL concurrency test)
bunx tsc --noEmit # type-check
Built test-first across 14 TDD tasks with per-task and whole-branch review.
License
MIT © 2026 Vorluno
<div align="center">
Built by Vorluno — a software studio from Panamá 🇵🇦
Part of the mcp-s family of MCP servers.
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.