Skip to content
Get packs

Guide · Open source

How to use Distilly: build a Person Profile Agent Skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

Distilly is an open-source Python project that takes material you provide, such as messages, documents, interviews or public sources, and builds a Person Profile describing a person's experience, decision patterns, expression and ways of working. This guide explains what the project does, how to get the code, and how the output is packaged for AI agent hosts. By the end you will know what inputs work, what to expect from the result, and where the project's limits are.

Distillytitanwings/distilly · ★ 26.3k · MIT

Quick start

  1. 1Check Python. Confirm you have Python 3.9 or newer installed, as listed in the README prerequisites.
  2. 2Clone the repository. Run git clone https://github.com/titanwings/distilly to download the project.
  3. 3Gather source material. Collect messages, documents, interviews or public sources that show how the person works and communicates.
  4. 4Pick a character family. Choose colleague, relationship or celebrity, depending on the person and the material you have.
  5. 5Generate the Person Profile. Use the distilly creator to build the profile from your materials, following the project README.
  6. 6Install the Skill in a host. Place the generated Skill in a supported agent host such as Claude Code or Codex and invoke it.

What Distilly produces

Distilly builds a Person Profile from the material you supply. The README describes it as a source-grounded profile built from observable experience, decision patterns, expression and ways of working.

The profile is packaged as an Agent Skill. Supported agent hosts can install and invoke that Skill. The README says the project models observable patterns and does not claim to clone the person.

  • Inputs: messages, documents, interviews or public sources
  • Output: a Person Profile, packaged as an Agent Skill
  • Licence: MIT
  • Language: Python

Install Distilly by cloning the repository

Distilly uses a clone-and-build install. The README lists Python 3.9+ as the prerequisite. Clone the repository with the command below, then follow the build steps in the project's README.

The README names the canonical creator Skill as distilly, so install it in a directory named distilly.

  • Prerequisite: Python 3.9 or newer
git clone https://github.com/titanwings/distilly

Choose the right character family

Distilly organises its work into three families. Each has its own source-collection strategy, analysis dimensions and Person Profile structure.

The colleague family covers coworkers, mentors and teammates, and the README says it supports Lark, DingTalk and Slack collection. The relationship family covers exes, partners, parents, friends and close family. The celebrity family covers public figures, creators and fictional characters, and includes a six-dimension research toolchain.

  • Colleague: builds a Work Skill and Persona from workflows, expression and workplace behaviour
  • Relationship: organises expression patterns, emotional triggers, conflict and repair patterns into a Persona Skill
  • Celebrity: uses subtitles, transcript cleanup, research merge and quality check

Prepare good source material

The README states that output quality depends on the materials provided. If you have no source material, Distilly is not a good fit.

Use material that shows how the person actually works and communicates. The more representative the material, the more grounded the profile is likely to be.

  • Messages, documents, interviews or public sources
  • Material written or said by the person themselves
  • Enough volume to show repeated patterns rather than one-off remarks

Use the Skill in a supported agent host

The current release supports native local Skill discovery for several hosts, including Claude Code, Codex, Hermes, OpenClaw, DeepSeek Harness, Pi, Grok Build and OpenCode. The README says Grok Bot supports saved or private Skills but does not describe direct local SKILL.md imports, so direct repo installation there is not yet verified.

Once installed in a supported host, the agent can invoke the Person Profile Skill to answer in that person's style and draw on their working knowledge.

  • Claude Code, Codex, Hermes, OpenClaw, DeepSeek Harness, Pi, Grok Build, OpenCode are listed as supported hosts
  • Grok Bot: manual migration into a saved Skill only, per the README

What people use it for

  • A developer gives an agent the working style of a senior colleague to answer code review questions
  • A team preserves a departing colleague's working knowledge as a reusable Skill
  • A user builds a profile of an author from public writing and interviews
  • A mentee asks an agent to respond the way a mentor would on career decisions

Frequently asked questions

Is Distilly free to use?

Distilly is released under the MIT licence, which is a permissive open-source licence. Check the LICENSE file in the repository for the full terms before you use or redistribute it.

What do I need to run Distilly?

The README lists Python 3.9 or newer as the prerequisite and uses a clone-and-build install. You also need source material such as messages, documents, interviews or public sources, because output quality depends on it.

How do I install Distilly?

Clone the repository with git clone https://github.com/titanwings/distilly, then follow the build steps in the README. The creator Skill is named distilly and should be installed in a directory of that name.

Can Distilly clone a real person?

No. The README says Distilly models observable patterns and does not claim to clone the person behind the material. It is not a verified clone of anyone.

Is there a one-click app for Distilly?

The README describes a Skill-based workflow for agent hosts rather than a consumer app. It is aimed at people comfortable cloning a repository and installing Skills.

Are there alternatives or comparable projects to Distilly?

The project's analysis lists no comparable projects. Distilly was formerly called Colleague Skill, which the README notes for search continuity.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Distilly suits developers and teams who want an agent to answer in the style of a specific colleague, mentor or public figure, using material they supply and install as an Agent Skill on a supported host. It is a poor choice if you need a verified clone of a real person, have no source material, or want a one-click consumer app.

Install options & downloads