session:join

Join multi-agent harness sessions and register an agent with a role and runtime.

Updated Mar 14, 2026
One-click install
npx skills add https://github.com/smykla-skalski/harness --skill session-join
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: session:join
Source: https://github.com/smykla-skalski/harness/tree/main/plugins/harness/skills/join
Command: npx skills add https://github.com/smykla-skalski/harness --skill session-join

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Joining and participating in coordinated multi-agent harness sessions can be confusing without clear role semantics, session discovery, and standardized CLI steps; this Skill codifies the join workflow so agents can reliably register, claim work, observe, and report progress without touching orchestration files directly.

Core Features & Use Cases

  • Session discovery and picker: List active sessions, prompt the user when needed, and prefer the most recently created session for selection.
  • Role-aware join flow: Join as worker, observer, reviewer, or improver and follow role-specific workflows for claiming tasks, running observations, or performing reviews.
  • Operational guardrails: Enforce use of harness session commands, require --actor for mutating operations, and document task checkpointing, assignment, and completion patterns.
  • Use Case: A developer agent joins a test coordination session, claims tasks, runs checks and tests, records checkpoints, and marks tasks done while observers monitor and create triage tasks from findings.

Quick Start

Use the session:join skill to join a session as a worker, let the runtime auto-detect, and follow the CLI output to obtain your agent ID for --actor.

Frequently Asked Questions about session:join

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I join a multi-agent development session and register my role?

The session:join Skill streamlines multi-agent session registration by listing active sessions and prompting you to select one, preferring the most recently created. It then assigns an agent ID for --actor to participate in coordinated test workflows.

Can I join a coordination session as an observer instead of a worker?

Yes, you can join as a worker, observer, reviewer, or improver. Role-aware join flows dictate your permissions, allowing observers to monitor progress and create triage tasks from findings while workers claim and complete assigned tasks.

Does this approach support multiple AI coding runtimes like Claude and Copilot?

Yes, multi-agent session coordination supports claude, codex, gemini, copilot, and opencode runtimes. The harness CLI standardizes task management, signals, and observation pipelines across these diverse agent environments.

What's the best way to claim and report progress on tasks in a tracked session?

The best way to manage tasks is using role-specific flows via the harness CLI, requiring the --actor flag for mutations. Workers claim tasks, run checks, record checkpoints, and mark tasks done, ensuring progress is tracked.

Do I need to manually edit orchestration files to participate in agent workflows?

No, you do not need to touch orchestration files directly. The session join workflow codifies CLI steps for registration, task claiming, and reporting, enforcing operational guardrails through harness session commands instead of manual edits.

Why does the system require an agent ID for mutating operations during a session?

The system requires an agent ID via the --actor flag for mutating operations to enforce operational guardrails. This ensures accountability when claiming tasks, recording checkpoints, or marking work done across multi-agent sessions.