# Co Supervision

> Configure the project's supervision policy in RESEARCH.md. Uses a preset-first flow (`manual`, `checkpointed`, `autonomous`, `wild`), then lets the user adjust notification events, approval gates, stop limits, resource rules, and idea-change handling. Trigger phrases: "configure supervision", "set supervision", "automation settings", "change autonomy", "/supervision".

- Skill: `zy-ning/co-supervision` (Agent Skill)
- Install (CLI): `npx skillmds@latest add zy-ning/co-supervision`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zy-ning/co-supervision/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: zy-ning (https://skillmd.com/u/zy-ning)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zy-ning/co-supervision

---


# Co-Supervision

Use this skill when the user wants to configure how much autonomy the agent should use for the current project.

## Scope

This skill manages `## Supervision Policy` in `RESEARCH.md`.

If the user wants to configure both supervision and preferred skillpacks together, prefer `co-customize` and keep this skill focused on supervision only.

## Flow

Use a preset-first flow:

1. choose a preset: `manual`, `checkpointed`, `autonomous`, or `wild`
2. confirm notification events
3. confirm approval gates
4. confirm stop target / limits
5. confirm resource rules
6. confirm idea-change handling
7. write the policy into `RESEARCH.md`

## Precedence

`RESEARCH.md` `## Supervision Policy` is the live working-state layer and overrides everything.
`.co-researcher/skills.yaml` `supervision` is the project-default layer written by `co-customize` — only consulted when `RESEARCH.md` has no policy yet.

## Constraints

- Do not create a separate hidden config store for supervision.
- Keep the policy human-readable.
- If `RESEARCH.md` is missing, initialize it before editing.
- `wild` is never implicit; require explicit user confirmation.