alethe-orchestrate skill
Hands routine work to Codex, Claude Code, Antigravity and Kimi workers via playbooks and reports what each did and cost.
- 142 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Launches other CLI agents that edit files in worker worktrees without your involvement
- Spends quota and money across several subscriptions and APIs at once
- The hook reads local session logs on every prompt
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/alethe-orchestrateDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Launches other CLI agents that edit files in worker worktrees without your involvement
- Spends quota and money across several subscriptions and APIs at once
- The hook reads local session logs on every prompt
This is third-party code. Review the repository files before installing.
What it does
alethe-orchestrate turns the main agent into a lead: it picks a playbook, writes briefs, shows the plan and runs workers through other agents' CLIs. Workers receive only their part of the task and leave `result.md`, `diff.patch` and `outcome.json`, while full transcripts stay on disk. The lead reads only the results. There are eleven playbooks, from `direct` and `cascade` to `parallel-worktrees`, `best-of-n` and `review-gate`. The `/alethe report` command shows token use and cost per worker. Passive mode watches session spend and itself suggests handing small tasks to a cheaper model.
Who it is for. People running large edits across several coding-agent subscriptions who want to save the lead model's quota.
Good fit when
- A migration of 40 endpoints should be split between Codex and Claude Code in separate worktrees.
- Cheap scouts first read unfamiliar code and the lead receives only summaries.
- A finished diff is reviewed before merge without spending the most expensive model on it.
- Claude quota is running out and small mechanical edits should go to Codex.
Not a fit when
- The edit touches one or two files and orchestration overhead outweighs the gain.
- Parts of the task are tightly coupled and need one context.
- Other agents' CLIs are not installed or not logged in.
Example request
`/alethe` move the billing endpoints to v2: scouting first, then edits in separate worktrees, then a diff review.Limitations
The skill launches other agents' CLIs, so installed and logged-in `claude`, `codex` or `agy` are required. Writing workers get their own worktree, and permissions and budget limits come from the user policy in the `trust` block. Cost is estimated at list prices as a subscription yardstick, not a bill. For Kimi and custom CLIs the read-only mode is checked afterwards through git. The Claude Code hook reads local session logs and Alethe state on every prompt and, according to the README, sends nothing over the network.
How to disable. `/alethe off` turns the mode off in a project and `/alethe off everywhere` everywhere. To remove the hook, uninstall the plugin or the `skills/alethe` folder, and `/alethe clean --yes` deletes old worker folders.
FAQ
Which agents can be workers?
Claude Code, Codex, Antigravity and Kimi Code. Any other CLI with a non-interactive mode joins as a custom worker through an adapter.
How do I install it as a plugin?
In Claude Code: `/plugin marketplace add Kc1t/alethe-orchestrate-skill`, then `/plugin install alethe-orchestrate@alethe`. Codex has the `codex plugin marketplace add` and `codex plugin add` pair.
What does passive mode do?
It watches token spend and quota and hands suitable tasks to workers without picking a playbook. If quota is unknown or the worker is billed by API, it asks for consent.
How do I see what each worker did?
`/alethe report` shows the work, time, tokens and cost, and `/alethe report html` opens the same as a page.
How do I stop running workers?
`/alethe stop` halts the workers and `/alethe status` shows where the run stands.
Discussion
No questions or reviews yet.
Related
SkillsOne agent launches others in the background: Claude Code, Codex, Copilot CLI, Grok Build and Pi, with a model chosen per stage.
A planner for Claude Code, Codex and OpenCode: one goal becomes a task plan and independent tasks run in parallel
A layer over coding agents that assembles a temporary specialist team per task inside Claude Code, Codex, Cursor and Gemini
A skill that turns the main Claude Code session into a dispatcher: tasks go to background subagents and the chat never blocks.
multi-agents runs a team of coding agents on a task and measures whether it beat a single agent and what it cost.
A skill with 25 commands to audit and improve the prompts, tools and agent designs of a project in Cursor, Claude Code and Codex
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| alethe-orchestrate (this card) | no | 142 | |
| OpenCode Agent Orchestration Kit | no | 153 | opencode-agent-orchestration-kit builds orchestration inside OpenCode; alethe runs CLIs from different vendors and merges spend into one report |
| Claw Orchestrator | no | 606 | claw-orchestrator manages agent sessions through a CLI; alethe adds playbooks and quota tracking |
README in short
Token and cost figures come from the CLIs' own reports rather than guesses. Before each wave the skill reads remaining quota and asks consent when the limit is close.