# Manage Local CLI Agents with tmux

Coordinating several local CLI agents manually makes task assignment, monitoring, and scheduling difficult. This skill provides tmux-based commands for starting, assigning, and observing agents.

## Install

```bash
npx skillstore add sickn33/agent-manager-skill
```

## Metadata

- Status: approved
- Slug: sickn33-agent-manager-skill
- Skillstore revision: r2
- Version status: missing
- Tree hash: 7dff54f7d1b023775e30a4e559fde1456891eda5b294aea02771434decf3f014
- Author: sickn33
- GitHub username: sickn33
- License: MIT
- Repository: https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/agent-manager-skill
- Ref: ebdfe608f5de2b66ff37ab4af12af8ac4f5e8006
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/sickn33-agent-manager-skill
- Manifest: https://skillstore.pages.dev/api/skills/sickn33-agent-manager-skill/manifest

## Capabilities

- Explains how to run multiple local CLI agents in separate tmux sessions.
- Provides commands to check the environment and list configured agents.
- Provides commands to start an agent and follow its monitor output.
- Shows how to assign a workflow to a named agent through standard input.
- Identifies the agents directory and supports cron-oriented recurring work.

## Use Cases

- Parallel Development Agents: Run independent coding tasks across separate local sessions while keeping each agent output isolated.
- Routine Repository Maintenance: Assign recurring maintenance workflows and prepare controlled cron schedules for local repositories.
- Agent Fleet Observation: List agents and follow session output during coordinated multi-agent work.

## Prompt Templates

### Check Prerequisites

```
Run the agent-manager doctor check. Report whether tmux, Python, and agent configuration are ready. Do not change anything.
```

### List and Inspect Agents

```
List configured agents, summarize their current states, and identify any agent needing attention. Do not start or stop sessions.
```

### Start and Monitor an Agent

```
Start agent [AGENT_ID], then monitor its output. Confirm the session started and summarize errors without changing other agents.
```

### Coordinate a Scheduled Workflow

```
Review [WORKFLOW_FILE], assign it to [AGENT_ID], and propose a cron schedule. Ask before creating cron entries or modifying agent configuration.
```

## Limitations

- Requires tmux, Python 3, and the separately cloned agent-manager repository.
- Includes guidance only and does not bundle the referenced management scripts.
- Does not document access controls, resource limits, failure recovery, or session isolation.
- Requires environment-specific validation before scheduled or consequential operations.

## Best Practices

- Run the doctor check before starting agents or assigning workflows.
- Use explicit agent identifiers and review current state before every action.
- Review workflows and obtain approval before cron or consequential operations.

## Anti Patterns

- Do not include credentials or sensitive data in assignments or monitored output.
- Do not start duplicate sessions without checking the current agent list.
- Do not treat monitor output as proof that a workflow completed correctly.

## Security Audit

- Audited at: 2026-07-23T20:13:31.747\+00:00
- Summary: Six external-command findings are false positives caused by Markdown inline-code and fenced-code syntax; the document contains no shell backtick execution. The GitHub clone instruction is a confirmed low-severity supply-chain risk because it fetches mutable, unverified external content. No prompt injection or data-exfiltration intent was found.

## Stats

- Views: 161
- Downloads: 10
- Favorites: 0
- Popularity score: 0