Skip to content
SkillScript-led852 tokens in use

Distilly

Distills a colleague, relationship or public figure from messages, documents and interviews into a reusable Person Profile, and sets up an owner profile for your agent.

26,706 stars on the repository MIT Licensetitanwings/distilly
git clone https://github.com/titanwings/distilly <DISTILLY_SKILL_DIR>
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Distilly costs

46tokens idle
852tokens in use
9,548tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Script-led

Script-ledMixedModel-led

Most of this skill is scripts, so it behaves much the same whichever model runs it.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: bin/distilly.mjs, tools/credentials.py, tools/dingtalk_auto_collector.py, tools/email_parser.py, tools/feishu_auto_collector.py, tools/feishu_browser.py, tools/feishu_mcp_client.py, tools/feishu_parser.py, tools/hosts.py, tools/install_claude_generated_skill.py, tools/install_distilly.py, tools/install_generated_skill.py, tools/install_generated_skill_common.py, tools/owner_profile.py, tools/owner_sources.py, tools/research/__init__.py, tools/research/download_subtitles.sh, tools/research/merge_research.py, tools/research/quality_check.py, tools/research/srt_to_transcript.py, tools/research/transcribe_audio.py, tools/research/xquik_public_posts.py, tools/skill_presets.py, tools/skill_schema.py, tools/skill_writer.py, tools/slack_auto_collector.py, tools/version_manager.py.

Touches

shelldeclared and detectednetworkdetectedwrites filesdeclared and detectedcredentialsdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Fair
  • Frontmatter8/15
  • Structure11/25
  • Examples3/15
  • Output10/15
  • Size15/15
  • Resources12/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Needs attention
  • File that was not read−15

    a file the skill ships that was not read, so what it does is unknown

    .npmrc
  • File that was not read−5

    a file the skill ships that was not read, so what it does is unknown

    CITATION.cff
  • Hidden characters−20

    characters that render as nothing or reorder the text

    README.md, line 82
  • Hidden characters−20

    characters that render as nothing or reorder the text

    README.md, line 86
  • Hidden characters−20

    characters that render as nothing or reorder the text

    README.md, line 112
  • Reaches the network−10

    does this on the machine, and the frontmatter does not say so

    tools/credentials.py, line 172
  • Writes files−4

    does this on the machine, as the frontmatter declares

    tools/dingtalk_auto_collector.py, line 69
  • Runs shell commands−2

    does this on the machine, as the frontmatter declares

    tools/feishu_mcp_client.py, line 129
  • Reads credentials−15

    does this on the machine, and the frontmatter does not say so

    tools/research/transcribe_audio.py, line 145

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity76
  • Maintenance68
  • Community52
  • Maturity28
  • Momentum40

Distilly (formerly Colleague Skill) builds source-grounded Person Profiles for agents. You pick one of three families, colleague, relationship or celebrity, add source material, review the result, and the skill packages the profile as its own Agent Skill.

It also has an owner flow: a short interview plus the sources your agent already has, saved as a profile and linked into the file your agent reads every session. Generated profiles can be updated with new files, corrected in conversation and rolled back.

What You Get

  • A generated Agent Skill per person, stored under skills/colleague, skills/relationship or skills/celebrity in your workspace
  • An owner profile saved locally and linked into your agent’s always-on user file
  • Append, correct and roll back commands for keeping a profile current
  • Collectors that pull messages and documents from Lark, DingTalk and Slack
  • A celebrity research pipeline from subtitles to a quality-checked dossier

Common Use Cases

  • Keep a departing colleague’s workflow and review style available to the team’s agent
  • Build a profile of a public thinker from books and interviews to ask how they would answer
  • Tell a new personal agent who you are with a five-minute interview
  • Parse exported chat history or email into a profile of a long-time contact

What Makes It Unique

Most persona skills ask for a hand-written description. Distilly starts from source material, ranks which sources count most for each family, and states that the output is grounded in what was observed and does not claim to clone the person. It keeps credentials out of the conversation by having you run setup commands in your own terminal, and it versions every update so a profile can be rolled back.

Under The Hood

SKILL.md lists allowed tools Read, Write, Edit and Bash and routes to prompt files for each flow. The agent runs bundled Python tools through the shell: a skill writer that creates and updates generated skills, parsers for Lark exports and email, and auto-collectors that call the Lark, DingTalk and Slack APIs and a browser session with credentials you set up yourself, stored in the OS keychain or a config file in your home folder. The owner flow looks at files and connector configs on your machine, which you choose from a list, and edits only the marked block of your agent’s user file after showing a dry run. It writes generated skills and knowledge folders in your project, keeps profiles under a folder in your home directory, and can send requests to public sources and a third-party service for celebrity research.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers