Endpoint AI

Import Claude session prior art without installing Harriet Desktop

Capture experiments from personal ChatGPT, Gemini, or Claude chats into Harriet’s skill review queue using the Session distiller skill and a short-lived upload key—no agent enrollment or MCP deploy first.

Use this flow when someone has already built something useful in a personal AI chat (ChatGPT, Gemini, or Claude — a widget, a repeatable workflow, or ad-hoc tool use) and you want it reviewed and stored in Harriet before rolling it out to the team.

You do not need Harriet Desktop, a background agent, or deployed MCPs just to submit the draft.

1. Download Session distiller

In Endpoint AI (Provisioner):

  1. Open Help → Session distiller (or My AI → Advanced options for a link).
  2. Click Download Session distiller (zip).
  3. Add the zip to Claude (Projects or your Claude skill workflow), or follow the skill’s ChatGPT instructions.

2. Distill the session in your chat app

In the same session where the experiment lives, load Session distiller and follow its instructions. It will help you produce:

  • A SKILL.md file (instructions for a repeatable skill)
  • Optional references/ markdown files

It will not ask you to host HTML dashboards or widget UIs on Harriet—those become written procedures instead.

3. Generate an upload key in Harriet

Still in Provisioner on the Session distiller Help tab:

  1. Click Generate upload key.
  2. Copy server, token (hss_…), and expiry into your chat app when prompted.

The token is shown once and expires after about an hour. You can revoke keys from the same screen.

4. Submit for review

Session distiller uses Harriet’s Skilify submit API with your upload key. After a successful submit, the package appears in the review queue (same as other package skills). It is not published to teammates until an approver acts.

Confirm with the author that reviewers may see the package before submitting.

5. After approval

Administrators assign the approved skill and any MCP connectors through normal Endpoint AI provisioning. See Using Skilify to build skills locally, submit for review, and provision them for the full author → review → deploy loop.

Related articles

  • Using Skilify to build skills locally, submit for review, and provision them — packaging from a provisioned endpoint
  • How do I share Harriet skills with my team? — assignments after approval

Use Harriet in your organisation for searchable help, AI assistance, and your company knowledge base.

Log in to Harriet