Didi Mind

An agent-first personal knowledge OS

Let agents read, write and maintain your knowledge base; you review and approve by conversation.
Everything carries provenance — AI never invents on your behalf.

didi-mind — sessionProvenance on

$ didi

didi-mind v0.1 · an agent-first personal knowledge OS

Knowledge base loaded: libs/ notes/ refs/ thoughts/

Provenance gate on · lint gate on

You Newsflash: <URL>

✓ signal/2026-08-22 filed

You Log an idea / my business direction

✓ thoughts/internalize.md updated

You

Your knowledge,
kept by agents

"AI needs a bibliography" — the risk of AI is, at bottom, an attribution problem.
Didi Mind turns that paper into a personal knowledge system:
every entry must name its source, with numbered citations, report audits and a lint gate, so AI output stays traceable.

Knowledge base — yours
Rare datalibs/ + notes/

The patent material, business topics and personal notes you have accumulated — AI organises it, but the content stays yours.

Tasterefs/

The references and standards you believe in, so AI holds the line on your judgement.

Trustthoughts/ + community

Your thinking and your running list of topics, plus the trust a community accumulates.

Just say it to the agent

No syntax to learn: one sentence is enough for the agent to file your knowledge.

didi-mind — commandsInteractive

# pick a command and watch how it works

One-line install:
clone it and start using it

The repository ships with the author's example knowledge base (business topics, taste standards and so on). During install, didi init automatically clears the author's personal content and gives the installer a brand-new empty knowledge base.

  1. 01
    git --version && go version

    Check the environment

    Confirm git and go (>=1.25) are available; install anything missing first (macOS: brew install go).

  2. 02
    git clone https://github.com/amorist/didi-mind.git

    Clone the repository

    Clone the Didi Mind repository onto your machine.

  3. 03
    cd didi-mind && make build

    Build and pass the gates

    Gates: validate pipelines/templates → unit tests → compile the CLI → sync the release build.

  4. 04
    ./skills/scripts/didi init --yes

    Initialise an empty knowledge base

    Clears the author's example knowledge base (business topics, taste standards and so on) and hands the installer a brand-new empty one.

  5. 05
    brew install graphviz

    Optional: graph rendering

    Set up the graph rendering environment (optional).

  6. 06
    ./skills/scripts/didi pipeline list && didi template list && didi kb lint

    Verify the install

    pipeline list should show 10 pipelines, template list a/b/c/d/e and kb lint no errors — then you are ready to start.

  7. 07
    Read AGENTS.md and skills/SKILL.md

    Read the docs

    Read AGENTS.md and skills/SKILL.md, then report the install result and how to begin.

  8. 08
    Tell Didi: my identity / my business direction / my current goal

    Complete your personal profile

    Walks the installer through a personal profile: generate the libs/people/me/ owner-page skeleton and prompt for it (never invent the content).

View the source (GitHub)

Build your own
knowledge operating system

Go to GitHub
Didi Mind · AI WORKSHOP