ClaudeMap

Agenttrail

Local observability for AI coding agents — a live map of what changed, what's planned, and where your attention is needed.

SDKs & Toolingobservabilitymonitoringdevtoolsnpm
Visit projectsodiumsun/agenttrail

Agenttrail watches the activity your coding agents (Claude Code, Codex) already produce and turns it into two live views: Map renders the project's file tree, dependency links, and session trails across multiple repos, while Kitchen visualizes the same evidence as dishes moving through stations. It observes — it does not run agents or decide tasks are done. npx agenttrail init optionally writes conventions into CLAUDE.md/AGENTS.md, creates a starter plan, and offers local Claude Code hooks; normal watching writes nothing.

What it does

  • Two views over the same evidence: Map (file tree, dependency links, session trails) and Kitchen (stations and dishes)
  • Multi-repo overview with saved boards (npx agenttrail up) and login autostart
  • npx agenttrail init writes conventions to CLAUDE.md/AGENTS.md, adds a starter plan, offers Claude Code hooks — watching itself writes nothing
  • Fully local by construction; npm packages bundle graphics and fonts, no build step

When to use it

For developers running several long-lived agent sessions who want to see where work is happening without tailing N terminals.

Getting started

npx agenttrail            # Map: project structure & session trails
npx agenttrail-kitchen .  # Kitchen view for the current repo
npx agenttrail init       # one-time conventions + starter plan (writes files)

Things to know

Public experimental preview (0.1.0-alpha line) — pin the version if you want reproducibility. Agent support centers on Codex and Claude Code; other tools may show partial signal.

Repository

sodiumsun/agenttrail

Charted

Related on the map

  • Agent Skills (Open Standard)

    SDK / Tool

    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
    View

  • 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
    View

  • 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
    View