tomtomsatnav/agent-orchestration-standard/dashboard
Claude Code and agent skill by tomtomsatnav/agent-orchestration-standard
- community
Checking that the SKILL.md file is still in the repository.
The AOS dashboard: a read-only live monitor of Claude Code agents (agents, tasks, cost, file conflicts, alerts, history) served at http://127.0.0.1:4318. Use when the user asks about the AOS dashboard, live agent status, what agents are doing, session cost, file collisions between agents, or how to open or disable the monitor.
Install
npx vskill@latest install tomtomsatnav/agent-orchestration-standard/dashboardSKILL.md
AOS dashboard
The AOS dashboard is a local, read-only web view of Claude Code activity in AOS-1 form. The plugin's hooks log one compact line per hook event (session start and end, prompts, tool calls, subagents, stops, notifications) to a local file. A small monitor folds that log into an AOS-1 state view and serves it.
Open it
- URL: http://127.0.0.1:4318
- macOS:
open http://127.0.0.1:4318 - Terminal version:
node ${CLAUDE_PLUGIN_ROOT}/cli/dash.mjs
Autostart
A SessionStart hook (hooks/ensure-monitor.mjs) starts the monitor when it is not already answering. It is idempotent: it probes GET /api/state on 127.0.0.1 first, takes a lock so two sessions cannot start two monitors, waits up to 3 seconds for the monitor to come up, and always exits silently. The monitor binds 127.0.0.1 only.
To disable autostart, set AOS_NO_AUTOSTART=1 in the environment Claude Code starts in. Then start the monitor yourself if you want it, or leave it off.
The monitor port can be changed with AOS_PORT (or PORT) for the autostart check and the monitor.
Data and privacy
- Read-only. The dashboard and the MCP tools never change agent state.
- Local only. Nothing leaves the machine.
- Stored data lives in the plugin data directory (
${CLAUDE_PLUGIN_DATA}/live, orAOS_LIVE_DIRif set). - The hooks store no prompts, no command text, no file contents, no tool output and no environment values. They keep event names, tool names, session IDs (truncated), agent IDs and types, basenames of most tool targets, the first word of Bash commands (only when it is a plain word), the full path of files touched by Edit/Write tools, transcript paths, and a notification type.
MCP tools
The aos MCP server exposes read-only views of the same state:
aos_status: overall state of sessions and agentsaos_agent: one agent's activityaos_cost: token and cost estimates (dollar values need prices inprices.json)aos_conflicts: files touched by more than one agentaos_alerts: stuck or needs-input alertsaos_history: recent eventsaos_state_at: state at a past timestamp
Settings
live/config.json:stuck_after_sandnotifysettings. Defaults are copied in on first run and never overwritten.live/prices.json: per-model USD per million tokens. Left empty by default, so cost stays unknown until filled in.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugin/skills/dashboard/SKILL.md
Popularity
- Repository stars
- 70
- Repository forks
- 9
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/tomtomsatnav/agent-orchestration-standard/dashboard)