# Praxist Run Control

Starts, stops, resumes and inspects Praxist task runs from an agent interface.

- Repository: https://github.com/sapientinc/PRAXIST (5536 stars)
- Category: AI and agents
- License: see repository
- Last updated: 2026-09-23
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add sapientinc/PRAXIST --skill praxist-control -g -a claude-code -y`
Codex: `npx skills add sapientinc/PRAXIST --skill praxist-control -g -a codex -y`
Cursor: `npx skills add sapientinc/PRAXIST --skill praxist-control -g -a cursor -y`
Gemini CLI: `npx skills add sapientinc/PRAXIST --skill praxist-control -g -a gemini-cli -y`

## Description

Start, stop, resume, inspect status, open the independent read-only foreground TUI, detect active runs, and safely repair Praxist run lifecycle boundaries from a supported agent interface. Use when the user asks to launch a Praxist task, stop or kill a run, continue/resume an interrupted run, restart from the latest safe generation, query current run progress, open or exit a live monitor, detect or list currently running Praxist tasks in the environment, inspect generation status, view incubator/frontier/leaderboard performance, check hardware load, handle interrupted PI panel or Gems reset boundaries, inspect whether a task directory is runnable, or control Praxist lifecycle commands with `praxist start`, `praxist stop`, `praxist status`, `praxist --monitor`, `praxist resume`, or `praxist resolve`.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Praxist Control

Use this skill to operate an existing Praxist task project. It controls runs;
it does not create task projects, install dependencies, edit Praxist core, or change
provider credentials.

## Empty Request Guard

If the user invokes this skill without naming an operation, exit without taking
an action. Ask the user to specify one of: `start`, `stop`, `resume`, `status`,
`monitor`, or `detect-active-runs`. Do not infer a default lifecycle action from
an empty prompt.

## Command Surface

Verify the live CLI before acting:

```bash
praxist --help
praxist start --help
praxist stop --help
praxist status --help
praxist --monitor --help
praxist resume --help
praxist resolve --help
```

If `praxist` is not on `PATH` but the current directory is a Praxist source
checkout, use `uv run praxist ...`. If neither works, refuse the lifecycle action and
tell the user to install or activate Praxist first.

Prefer JSON for machine decisions:

```bash
praxist status --json
praxist start --task-path /path/to/task --daemonize --json
praxist stop <run_id> --grace 300 --json
praxist resume <run_id-or-run-dir> --daemonize --json
```

Use `--daemonize` for `start` and `resume` from agent or CI shells, or other
sandboxed shells so the run survives the launching process.

## Detect Active Runs Workflow

Use this read-only workflow when the user asks which Praxist tasks are currently
running, whether the environment is idle, whether another server/session has a
live run, or explicitly asks to "detect active run(s)".

1. Run:

   ```bash
   praxist status --json
   ```

The rest: https://github.com/sapientinc/PRAXIST/blob/HEAD/skills/praxist-control/SKILL.md