ClaudeMap

SITREP

One live console for every Claude Code and Codex session across machines — which are working, waiting on you, stuck or failed — with answering, cost estimates and a wall view (MIT, macOS/Linux, Node 24)

SDKs & Toolingmonitoringdashboardfleetclaude codecodex

SITREP is a self-hosted dashboard for a fleet of coding agents: every session on every machine shows as a card, list row or graph node with its plan and current step, its agents, model and effort, context fill, git state, cost, and what failed. Everything waiting on you — permissions, questions, plans — sits at the top, longest wait first, and you answer it from the dashboard or the terminal; the first answer wins. It estimates spend by session, model and machine, shows each provider's plan window before it runs out, keeps history of ended sessions, and offers a wall view plus browser notifications. It reads sessions via reporters and only ever sends your own answers, messages and commands — it never controls a terminal, never starts, stops or restarts a session, and stores no transcripts. Claude Code and Codex sessions can be answered from the board; Hermes sessions are shown, not answered.

What it does

  • Fleet board across machines: cards, list or graph, grouped by machine or project, filtered by state, machine, project or harness
  • Needs-you and Problems queues: what is waiting on you, and what is stuck, failed or colliding on a file
  • Answering from the dashboard: permissions, questions and plans, plus a message, /compact or /clear — the first answer wins
  • Cost and limits: estimated spend by session, model and machine, and each provider's plan window before it runs out (costs are marked "est.")
  • History of ended sessions by day with model, duration, cost and how they ended; wall view and browser notifications

When to use it

Use when you run coding agents on more than one machine and lose track of which session is waiting on you — SITREP answers permissions and plans from one page without hunting for the right terminal, and shows where money and context are going across the fleet.

Getting started

Server on a machine that stays on:

git clone https://github.com/IluvatarLabs/sitrep ~/sitrep && cd ~/sitrep
npm ci && npm run build
npx tsx scripts/install.ts --server --bind 0.0.0.0 --apply

It runs as a user service and writes an admin secret to ~/.sitrep/server.env. On each machine that runs sessions (commands come filled in from Machines → Add a machine):

git clone https://github.com/IluvatarLabs/sitrep ~/sitrep && cd ~/sitrep && npm ci
npm run build -w codex-reporter                               # only where Codex runs
npx tsx scripts/install.ts --claude --codex --url http://<server IP>:7487 --token-stdin --apply

npx tsx scripts/doctor.ts checks a machine; scripts/uninstall.ts lists, then removes with --apply. Your own hooks are never touched.

Things to know

The README's own warning: SITREP is built for one person on a trusted LAN and speaks plain HTTP — anyone who can reach port 7487 can read every session, and answering requires signing in with the admin secret; the Operations doc explains how to narrow this. Requires git and Node 24. The codex-reporter build step is only needed where Codex runs, and Hermes sessions are view-only. A value a harness does not report shows as "—", never as a guess.

Repository

IluvatarLabs/sitrep

Charted

Related on the map

  • The open Agent Skills standard — SKILL.md folders any agent can load, governed by the Agentic AI Foundation since 2026-10-22.

    • official
    • spec
    • agent-skills
    • standard
  • Harness Bridge

    SDK / Tool

    A local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.

    • launcher
    • local-models
    • cli
    • menu-bar
  • Claude for Siri

    SDK / Tool

    Answers macOS Spotlight and Siri prompts with your signed-in Claude Code account — an experimental macOS 27 model-delegation provider.

    • macos
    • siri
    • spotlight
    • model-delegation