- Home
- Skills
- Agents & Orchestration
- ahub control
Works with the AI Agents you already use
ahub control
Manage, resume, and orchestrate Codex and Claude Code sessions via a local Agent Hub CLI.
Free
ahub control
Example session with this skill installed
Use ahub-control to create a Claude Code session for the /Users/dev/project directory and send the task described in task.txt.
- Read your context and instructions
- Compiled the ahub control
I have successfully initialized a new Claude Code session and dispatched the task. The session is currently active and being monitored for progress.
Session Execution Summary
| Field | Value |
|---|---|
| Session ID | s_8f2k93zh |
| Agent | claude-code |
| Project | /Users/dev/project |
| Permission Mode | auto |
| Status | connected / working |
Terminal Diagnostics
The session was created using ahub-cli session create. I have piped the contents of task.txt into the session using the --stdin flag to ensure the full context was delivered without shell escaping issues. The agent has transitioned to the working phase and is currently performing initial directory indexing.
Observed Activity
14:02:11- Session created with IDs_8f2k93zh.14:02:12- Task delivered via stdin.14:02:15- Phase changed:update(Indexing project files...).
Next steps
- Run
ahub-controlto inspect sessions_8f2k93zhand check for any pending Plan approvals. - Use
terminal s_8f2k93zhif the agent appears stalled to verify if a sudo or git credential prompt is visible. - Call
exit s_8f2k93zhonce the deliverables intask.txtare verified.
Connects securely to your tools. The creator never sees your data.
About this skill
Managing multi-agent sessions across complex local environments often leads to context drift, abandoned terminal processes, and fragmented visibility. This skill provides a structured interface to Agent Hub, allowing you to orchestrate Codex and Claude Code sessions through a unified CLI layer with precise control over lifecycle and state.
What it does
- Session management handles the creation, resumption, and termination of Codex and Claude Code sessions within specific project scopes.
- Workflow orchestration allows sending complex tasks via stdin to avoid shell interpolation errors and monitoring execution progress in real-time.
- State inspection retrieves granular session data including connection status, idle/working phases, and response history using NDJSON following.
- Interactive handling manages agent questions and Plan approvals using simulated key inputs or batch-sent answers.
- Environment resolution maps non-absolute project paths to verified directories and resolves agent nicknames to ensure correct runtime execution.
How it works
- Initialize by verifying
ahub-cliavailability and resolving the target project path and agent nickname through the Hub's local service. - Execute by creating a new session or resuming an existing ID with a defined permission mode such as
autooryolo. - Monitor by following session output from a specific baseline to capture assistant replies, progress updates, and input requests.
- Interact by responding to terminal-based prompts via the
keyscommand or interrupting stalled background processes to recalibrate. - Verify deliverables against the task brief before explicitly exiting or deleting the session history.
Frameworks & tools
This skill requires a local installation of Agent Hub and ahub-cli. It interfaces with Codex and Claude Code agents operating within Node.js environments.
Why this beats prompting it yourself
Raw prompting cannot maintain the state machine required to monitor long-running terminal sessions or handle the nuances of hasMore pagination in NDJSON streams. This skill provides the logic to safely navigate connection states and permission prompts that typically break standard agent-to-agent interactions.
Use cases
- Long-running refactors where a sub-agent needs to work autonomously inside a persistent Claude Code session over several hours.
- Multi-project coordination by switching between isolated agent sessions while maintaining unique working directories and worktrees.
- Automated approval flows using the
autopermission mode to bypass manual interventions for known safe operations.
Known limitations
Requires Agent Hub to be pre-installed and running locally. The skill does not support starting the Hub service itself. Commands are limited by CLI timeouts, typically between 15 to 45 seconds.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Requires Agent Hub and ahub-cli to be pre-installed and running. Does not support starting the Hub service. Execution is bound by CLI-specific timeouts.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 4 days ago
- Free with an Agensi account