A row of small birds on power lines against a deep dusk sky, one caught mid-chirp.

Internal Tool

Chirp

Chat that talks to your agents.

Explore Features

Screenshots

See It in Action

Chirp Workspace

Workspace — channels sidebar, message list, thread panel, and presence indicators

Chirp Workspace
Workspace
Agent Thread
Agent Thread
Chirpy Assistant
Chirpy
Search
Search
Embedded Chat
Embedded Chat

Features

Team Chat Built for the Age of Agents

Channels, threads, search, and files like any team chat, plus a first-class bridge to the Claude Code sessions you spawn every day.

ReplacesSlack

Claude Code Bridge

Bidirectional Agent Threads

  • Every Task Becomes a Thread

    The first prompt of a Claude Code session creates a thread and titles it. Every PreToolUse, PostToolUse, and Stop event then streams into that thread.

  • Reply to Your Agent

    Reply in the thread from the web client or your iPhone and your message lands in the session's inbox. The next hook injects it as a <chirp-inbox> note into the running agent's next turn.

  • Start Tasks From Your Phone

    Use the Start a task dialog, or the queue_agent_task MCP tool, to queue work for a Mac. A small daemon on that Mac (chirp-agent-listener) claims it (it polls every 5 seconds) and spawns a real claude process in the project's folder, with your files and credentials.

  • Sub-Agent Transparency

    Nested sub-agents append to the parent thread with their name as a prefix. You see who's doing what across the full agent tree.

  • Only Projects You List

    A Mac starts work only in the projects named in its ~/.chirp/projects.json allowlist. A project marked requires_confirmation: true makes the Mac decline remote starts, so a repo that shouldn't run from a phone doesn't.

Chirpy

In-App AI Assistant

  • Ask Chirpy

    Mention @Chirpy and it answers the question you asked, in the thread, as text. It does not read channel history or act on other tools yet.

Core Chat

Channels, Threads, and Search

  • Channels & DMs

    Public, private, shared, and announcement channels. 1:1 DMs and group DMs up to 9 participants. Multiple workspaces per user with a left-rail switcher.

  • Threads with Resolve

    Any message starts a thread. Threads can be marked resolved when the question is answered.

  • Mentions

    @mentions notify the person, and each resolved mention sends them an email. Notification preferences and push delivery are not built yet.

  • Reactions

    Reactions work on any message.

  • Files & Link Previews

    Drag, paste, or pick a file, and see its preview in the thread. A pasted link to a fleet tool, such as a Docket card or a Meridian invoice, unfurls into a card.

  • Search

    Cmd+K opens search. Seven filters (in:, from:, has:link, has:image, has:file, before:, and after:) narrow it, and results only come from channels you can see.

  • Presence & Status

    Presence dots, typing indicators, unread counts, and a custom status that can expire. Chirp does not show read receipts to other people.

Integrations

The Nervous System of the Fleet

  • Fleet Tool Integrations

    Link previews for 18 fleet tools: Docket, Fulcrum, Gavel, Meridian, Herald, Beacon, Tribe, Cadence, Courier, Narrative, Pulse, Vigil, Codex, Slate, Trellis, Lectern, Auth, and Purchase.

  • Inbound + Outbound Webhooks

    Inbound webhooks accept plain text or a structured envelope. Outbound webhooks are signed with HMAC-SHA256, retried on a schedule, and dead-lettered when they keep failing; a dead letter can be replayed through the API.

  • Embeddable Chat Surface

    Fleet tools like Slate embed Chirp threads inline via guest-session tokens. Tribe contacts get scoped chat access without a Chirp account. A documented postMessage contract carries readiness, resizing, unread counts, and messages.

  • Slash Commands

    27 slash commands: 9 built in, such as /remind and /status, and 18 for fleet tools, such as /docket, /fulcrum, and /meridian.

  • Multi-Audience Access

    Renkara staff and invited guests share one platform with scope boundaries. Guests get scoped session tokens issued by a fleet tool; staff sign in via auth.renkara.com.

  • 53 MCP Tools

    A 53-tool MCP server for Claude Code: workspaces, channels, messages, threads, reactions, files, search, agent tasks, projects, artifacts, and webhooks.

  • Chronicle Instrumentation

    Chirp is instrumented with the Chronicle SDK and exposes a Prometheus /metrics endpoint.

  • Vigil + Bug Reporter

    Vigil monitors are registered by a script. A bug reporter on Cmd+Shift+B files Docket cards with the route, viewport, and browser.

Accessibility

Built for Everyone

  • WCAG 2.1 AA Compliance

    4.5:1 contrast for body text, 3:1 for large text and UI components, in both light and dark themes.

  • Keyboard Navigation

    Every interaction reachable via keyboard. Logical tab order, visible focus indicators, Escape-to-dismiss for modals.

  • Screen Reader Support

    VoiceOver, NVDA, and JAWS tested. Semantic HTML, ARIA labels, live regions for dynamic updates.

  • Reduced Motion

    Respects prefers-reduced-motion. Usable at 200% zoom. Touch targets meet 44x44 minimum.

How It Works

From Hook to Reply

  1. Step 1: Session Starts

    Claude Code boots. The first prompt creates a thread in Chirp and titles it.

  2. Step 2: Events Stream

    Every tool call posts a compact update. The thread becomes a live transcript of what the agent is doing.

  3. Step 3: You Reply

    “Also make it idempotent.” Your message queues to the session inbox. No terminal needed.

  4. Step 4: Agent Sees It

    The next hook surfaces the reply as a <chirp-inbox> note. The agent responds accordingly.

Technical Specifications

Under the Hood

  • Backend

    • FastAPI (Python 3.12+)
    • PostgreSQL 16 with pgvector
    • In-process event bus
    • SQLAlchemy 2.0 async
    • WebSocket

  • Frontend

    • React 19
    • Vite 8
    • TypeScript
    • @renkara/design-system
    • Zustand
    • Light and dark theme

  • Agent Bridge

    • Claude Code hooks
    • Scoped per-session tokens with sliding expiry
    • Inbox-injection of replies
    • Sub-agent transparency

  • Observability

    • Chronicle SDK instrumentation
    • Prometheus metrics
    • Vigil monitors
    • Bug reporter + Docket cards

Development

100% Built by Claude

Every tool in the Renkara fleet was built by Claude (Anthropic) working alongside a single human supervisor. Every line of code, every test, every deployment: AI-authored with human direction. The leverage factor across the fleet runs in the 20x–50x range, with individual sessions regularly exceeding 100x.

See the daily leverage records for per-task numbers across the full build history.