Absolute Init logo

Absolute Init

Community
dhruvjainin
absolute-init

One-time setup for absolute: interview how you want it to behave (output style, autonomy, TDD strictness, spec dir, families) + detect the stack once, then write `.absolute.config.json` (project, committed) and `~/.absolute/config.json` (user defaults + per-project overrides). Every other absolute-* command reads it instead of re-detecting; non-blocking — commands proceed without it and soft-recommend it. Triggers on "absolute init", "set up absolute", "initialize absolute", "configure absolute", "first-time setup", "remember my conventions for this repo".

Overview

Publisherdhruvjainin
Repositoryabsolute
Skill nameabsolute-init
Stars
223
Forks
30
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by dhruvjainin on GitHub. Read the source before you install it.

Installation

Install the Absolute Init AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/dhruvjainin/absolute/tree/main/skills/absolute-init
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/dhruvjainin/absolute.git /tmp/absolute
mkdir -p .claude/skills
cp -r /tmp/absolute/skills/absolute-init .claude/skills/absolute-init
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Absolute Init in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Absolute Init on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Absolute Init is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

"set up / initialize / configure absolute"). Start your first response with the ⚙️ emoji.

Absolute Init

One-time setup. Detect the project's real conventions, ask a few questions about how you want absolute to behave, then write that to JSON config the other ten commands read on every run. Result: commands stop re-detecting your stack from scratch and respect your preferences (output style, gating, TDD strictness) without being told each time.

This is the only command that writes config. It is non-destructive: an existing config is shown and updated, never blindly overwritten. It never commits.


When to use

  • First time using absolute in a repo — gives every command cached conventions + your preferences.
  • Conventions changed (new package manager, new test/lint scripts, branch rename) — re-run to refresh.
  • You want to change how absolute behaves globally (output style, autonomy) across projects.

Other commands run fine without init — they fall back to on-the-fly detection and emit a one-line suggestion to run it. init just makes them faster and tailored.


Key principles

  1. Codebase before questions. Detect everything detectable first. Only ask what the repo can't tell you (preferences, ambiguous choices).
  2. A few questions, not a grill. ~4-6 max, one at a time. This is setup, not design review.
  3. Non-destructive. Existing config → show it, confirm each change, merge — don't clobber.
  4. Never auto-commit. Write the project file; tell the user to commit it. (Commit policy itself is not configurable — absolute never commits.)
  5. Two levels, project wins. Project config is team-shared and authoritative; global is your personal default + per-project overrides.

Step 1 — DETECT

Auto-detect the stack using the Codebase Convention Detection table in references/work.md (package manager, language/runtime, test runner, linter/formatter, build, CI, available scripts). Resolve each to the project's own script form (npm test, make lint) so cached commands match CI — not raw tools.

Read the actual package.json scripts / Makefile targets to fill test, lint, typecheck, format, build. Detect the default branch with git symbolic-ref --short refs/remotes/origin/HEAD (fallback main). Anything you can't resolve confidently → leave it out and ask, or omit.

Also detect, from deps and marker files (omit the whole block if nothing is found):

  • conventions.format — formatter script: package.json format/fmt scripts, a Makefile format target, ruff format, or gofmt.
  • conventions.ui — from package.json deps: framework (react/vue/svelte/none), styling (tailwindcss → tailwind; css-modules; styled-components; else vanilla), iconLibrary (lucide-react, @heroicons/, @phosphor-icons/, react-icons…), componentLib (shadcn marker / @mui/material → mui / @chakra-ui → chakra / none), and tokensPath (a tokens/theme CSS or TS file if one exists). Omit if the repo has no UI deps.
  • conventions.docs — stack + dir via absolute-docs's Stack Detection marker-file table (source.config.ts/fumadocs, docusaurus.config.*, Starlight, mkdocs.yml, .vitepress/, Mintlify, else markdown). Omit if no docs are present.

Step 2 — RESOLVE existing config

Before asking anything, check for existing config (precedence below). If found, print a compact summary of current values and ask whether to update (default) or start fresh. Treat existing values as the defaults for the interview so the user can keep them with one keystroke.

Step 3 — INTERVIEW

Use AskUserQuestion for every preference question. Present options as structured choices with descriptions; mark the recommended/default option first with "(Recommended)" in its label. Ask questions one at a time — wait for each answer before advancing.

Questions (skip any the existing config already answers, unless user asked to reconfigure):

Q1 — Output style

question: "Output style?"
header: "Output style"
options:
  - label: "normal (Recommended)"
    description: "Full prose, explanations — good for onboarding or unfamiliar codebases"
  - label: "terse"
    description: "Compressed, minimal prose — matches caveman mode"

Q2 — Autonomy

question: "Autonomy level?"
header: "Autonomy"
options:
  - label: "gate-all (Recommended)"
    description: "Confirm before every change/wave — full control"
  - label: "auto-low-risk"
    description: "Auto-apply obviously-safe health waves, gate risky ones"

Q3 — TDD strictness

question: "TDD strictness?"
header: "TDD"
options:
  - label: "strict (Recommended)"
    description: "Test-first, red→green — failing test must exist before any code"
  - label: "pragmatic"
    description: "Tests required but not strictly written first"

Q4 — Spec / docs output dir Ask as free text: "Where should spec/work write design docs? (default: docs/plans)". Accept blank to keep default.

Q5 — Relevant families

question: "Which command families do you use?"
header: "Families"
multiSelect: true
options:
  - label: "build (Recommended)"
    description: "work, spec, ui, simplify, docs"
  - label: "health (Recommended)"
    description: "upgrade, audit, prune, debt, deflake"

Q6 — Confirm detected conventions Show the detected conventions block as a code block and ask: "Are these correct? Reply with any corrections or press enter to accept.". Accept blank to proceed.

Step 4 — WRITE

Write .absolute.config.json at the repo root (pretty-printed, 2-space). This is the committed, team-shared file. Then ask whether to also persist to the global file ~/.absolute/config.json:

  • defaults.preferences — your cross-project preference defaults.
  • projects["<absolute repo path>"] — a per-project override entry (machine-local, not committed).

Merge into any existing global file; never drop unrelated keys. After writing, print the file paths and remind the user to commit .absolute.config.json (you never commit).


Config schema

Both files share conventions + preferences; the global file wraps them.

.absolute.config.json (project, committed):

json
{
  "version": 1,
  "conventions": {
    "packageManager": "npm",
    "languages": ["typescript"],
    "test": "npm test",
    "lint": "npm run lint",
    "typecheck": "npm run typecheck",
    "format": "npm run format",
    "build": "npm run build",
    "ci": ".github/workflows/ci.yml",
    "defaultBranch": "main",
    "ui": {
      "framework": "react",
      "styling": "tailwind",
      "iconLibrary": "lucide-react",
      "componentLib": "shadcn",
      "tokensPath": "src/styles/tokens.css"
    },
    "docs": {
      "stack": "fumadocs",
      "dir": "docs"
    }
  },
  "preferences": {
    "outputStyle": "normal",
    "autonomy": "gate-all",
    "tdd": "strict",
    "specDir": "docs/plans",
    "boardTracking": "gitignored",
    "families": ["build", "health"],
    "health": {
      "protectedPaths": ["dist/**", "vendor/**", "**/*.generated.*"],
      "deflakeRuns": 20
    }
  }
}

The conventions.ui and conventions.docs blocks are omitted entirely when nothing is detected (no UI deps, no docs stack), keeping the file minimal.

~/.absolute/config.json (user/global):

json
{
  "version": 1,
  "defaults": {
    "preferences": {
      "outputStyle": "normal",
      "autonomy": "gate-all",
      "tdd": "strict",
      "specDir": "docs/plans",
      "families": ["build", "health"]
    }
  },
  "projects": {
    "/Users/you/dev/your-repo": {
      "conventions": { "packageManager": "pnpm", "test": "pnpm test" },
      "preferences": { "outputStyle": "terse" }
    }
  }
}

Field meanings:

FieldValuesEffect
conventions.*stringsCached stack/scripts every command runs through instead of re-detecting.
conventions.formatstringFormatter script — simplify/debt run it instead of re-detecting.
conventions.uiobjectCached framework/styling/iconLibrary/componentLib/tokensPath — ui adopts these instead of assuming Tailwind / guessing the icon library.
conventions.docsobjectCached docs stack + dir — docs skips marker-file re-detection.
preferences.outputStylenormal | terseVerbosity of command responses.
preferences.autonomygate-all | auto-low-riskWhether health commands auto-apply safe waves.
preferences.tddstrict | pragmaticwork's test-first rigor.
preferences.specDirpathWhere spec/work write design docs.
preferences.boardTrackinggitignored | git-trackedWhether work's .absolute-work/board.md is committed — set once instead of asked each intake.
preferences.families["build","health"] subsetTrims the no-arg menu to what you use.
preferences.health.protectedPathsglob[]Never-delete globs prune honors (generated/vendored code).
preferences.health.deflakeRunsintDefault N repeat-runs deflake uses to establish flakiness.

Precedence (how commands read config)

At the start of any command, resolve effective config by overlaying, highest wins:

  1. ./.absolute.config.json (project, committed)
  2. ~/.absolute/config.json → projects["<cwd absolute path>"]
  3. ~/.absolute/config.json → defaults

Shallow-merge conventions and preferences separately; merge the nested conventions.ui, conventions.docs, and preferences.health blocks one level deeper so a partial override doesn't drop sibling keys. If none exist, there is no config — the command soft-recommends init and uses on-the-fly detection.


Gotchas

  1. Overwriting a hand-tuned config. Always RESOLVE first; merge, don't clobber.
  2. Caching wrong commands. Cached test/lint that don't match CI poison every later command — verify each detected script actually runs.
  3. Committing the global file. ~/.absolute/config.json is machine-local; only .absolute.config.json is committed.
  4. Stale conventions. A renamed branch or swapped package manager silently misroutes commands — re-run init after stack changes.
  5. Treating init as a gate. It isn't. Commands proceed without it; init is an optimization, not a prerequisite.

Companion commands

  • /absolute work — the main consumer of cached conventions + tdd/autonomy/boardTracking prefs.
  • /absolute spec — reads conventions + specDir instead of re-detecting.
  • /absolute ui — reads conventions.ui (framework, styling, icon library, tokens path).
  • /absolute simplify — reads cached test/lint/format/typecheck for auto-verify.
  • /absolute docs — reads conventions.docs (stack, dir), skipping marker re-detection.
  • /absolute-upgrade|audit|prune|debt|deflake — health family reads conventions for DETECT, autonomy for gating, and preferences.health.* (prune protectedPaths, deflake deflakeRuns).
  • Re-run /absolute init anytime conventions or preferences change.

Frequently asked questions

What does the Absolute Init AI skill do?

One-time setup for absolute: interview how you want it to behave (output style, autonomy, TDD strictness, spec dir, families) + detect the stack once, then write `.absolute.config.json` (project, committed) and `~/.absolute/config.json` (user defaults + per-project overrides). Every other absolute-* command reads it instead of re-detecting; non-blocking — commands proceed without it and soft-recommend it. Triggers on "absolute init", "set up absolute", "initialize absolute", "configure absolute", "first-time setup", "remember my conventions for this repo".

Why use Absolute Init on TypingMind?

Because you install it once and use it with any model. Absolute Init is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Absolute Init in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/dhruvjainin/absolute/tree/main/skills/absolute-init. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Absolute Init?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Absolute Init?

As many as you like. As long as a model supports skills, you can use Absolute Init with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Absolute Init AI skill free?

Yes. It is published on GitHub by dhruvjainin under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇