# Ki Acquire Chatgpt

> Govern ChatGPT conversation capture: honest digest or full-thread coverage, routing from project and conversation evidence, and the capture end point recorded in the digest's README front matter. Use for ChatGPT capture or audit; `ki-acquire` owns the common contract.

- Skill: `knowledgeislands/ki-acquire-chatgpt` (Agent Skill, multi-file: 14 files)
- Install (CLI): `npx get-skill add knowledgeislands/ki-acquire-chatgpt`
- Raw SKILL.md: https://api.skillmd.com/api/skills/knowledgeislands/ki-acquire-chatgpt/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: knowledgeislands (https://skillmd.com/u/knowledgeislands)
- Updated: 2026-11-16
- Page: https://skillmd.com/skills/knowledgeislands/ki-acquire-chatgpt

---


# ChatGPT acquisition

Apply `ki-acquire` for the common contract, then [the ChatGPT acquisition standard](references/standards-chatgpt-acquisition.md). The generated [rubric](references/rubric.md) holds the review criteria; read [the sources](references/sources.md) only during REFRESH or when current provider capability matters.

A ChatGPT capture is a digest or full-thread snapshot written by the receiving repository's capture route, with its end point in the digest's README front matter. The how-to has moved out of the backpack: the Inbox's muninn (krisb/inbox, `packages/ki-muninn/`) checks one user-prepared readable capture with `chatgpt import` and writes it as a Knowledge Export Package with `chatgpt export`, and its README and the Inbox's ChatGPT guide own the commands and `[muninn.chatgpt]` settings. `ki acquire` in tools-ki ran this bridge until 5 November 2026; `ki muninn` will run it once tools-ki's extension mechanism, KI-TOOL-KI-CLI-176, lands. This skill keeps the provider method. Acquisition never mutates ChatGPT.

## Operating modes

For `help`, `-h`, `?` or no recognised mode, explain the skill, modes and off-ramps, then stop; offer a mode choice only in an interactive session.

### Mode AUDIT

Run the `ki-acquire` audit, then `ki repo audit --skill ki-acquire-chatgpt --repo <repo>`, and review supplied capture evidence against the rubric. Missing coverage or end-point evidence stays a named gap; do not contact ChatGPT merely to complete an audit.

### Mode CONFORM

Run AUDIT, then `ki repo conform --skill ki-acquire-chatgpt --repo <repo> --dry-run`. It may publish the generated rubric but cannot invent routes, content, end points or retirement authority.

### Mode EDUCATE

Explain digest versus full-thread coverage, the front-matter end point, linked segments, routing and the separate retirement gate.

### Mode REFRESH

Run only in the canonical `ki-agent-backpack`; from an installed copy, stop and redirect there. Re-read [the sources](references/sources.md), inspect only official capability documentation and privacy-minimised schemas, update the standard and rubric when supported behaviour changes, and record the review date.

## Off-ramps

- The Inbox's muninn owns execution.
- `ki-env-housekeeping-chatgpt` owns opaque local-store evidence, which is never conversation content.
- ChatGPT account administration, project renaming, archiving and deletion stay outside acquisition authority.

