Skip to content
Skills Library
AI & agents · Building & coding

initializer

/initializer

Adaptive initializer session (Phase 1 of the greenfield guide). Reads the spec plus the product and design docs, profiles the project, and generates the project's memory (feature_list.json, init.sh, claude-progress.txt, CLAUDE.md, hooks, verify/) WITHOUT implementing features. With --resync, reconciles a changed spec with feature_list.json. Use when starting a greenfield project after the spec is approved, or when the spec changed ("initialize the project", "run the initializer", "resync", "the spec changed").

Licence
Apache-2.0
Last repository change
08.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
8
Scripts
1
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 08.11.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What to check before installing
  • 1 scripts in this package. Review them before installing, as Claude may execute them.

Scripts: templates/init.sh

What it does and when to use it

Adaptive initializer session (Phase 1 of the greenfield guide). Reads the spec plus the product and design docs, profiles the project, and generates the project's memory (feature_list.json, init.sh, claude-progress.txt, CLAUDE.md, hooks, verify/) WITHOUT implementing features. With --resync, reconciles a changed spec with feature_list.json. Use when starting a greenfield project after the spec is approved, or when the spec changed ("initialize the project", "run the initializer", "resync", "the spec changed").

Triggers from the description

initialize the projectrun the initializerresyncthe spec changed

Arguments

/initializer [--resync]
Original author description

Adaptive initializer session (Phase 1 of the greenfield guide). Reads the spec plus the product and design docs, profiles the project, and generates the project's memory (feature_list.json, init.sh, claude-progress.txt, CLAUDE.md, hooks, verify/) WITHOUT implementing features. With --resync, reconciles a changed spec with feature_list.json. Use when starting a greenfield project after the spec is approved, or when the spec changed ("initialize the project", "run the initializer", "resync", "the spec changed").

How it works

  1. Inviolable rules
  2. Default mode (initialize)
  3. --resync mode (the spec changed)
  4. What this skill does NOT do
Original SKILL.md headings · 69 lines, 589 words
SKILL.md preview
---
name: initializer
description: Adaptive initializer session (Phase 1 of the greenfield guide). Reads the spec plus the product and design docs, profiles the project, and generates the project's memory (feature_list.json, init.sh, claude-progress.txt, CLAUDE.md, hooks, verify/) WITHOUT implementing features. With --resync, reconciles a changed spec with feature_list.json. Use when starting a greenfield project after the spec is approved, or when the spec changed ("initialize the project", "run the initializer", "resync", "the spec changed").
argument-hint: "[--resync]"
---

# /greenfield:initializer

Implements Anthropic's *initializer agent* pattern ("Effective harnesses for long-running agents"), adapted to the spec instead of assuming a web app with 200 features.

## Inviolable rules
1. **Do not implement any feature.** Only a minimal skeleton, verification tooling and memory files.
2. **Do not commit until the human approves** the profile and the list.
3. Explain everything to the human **in plain language**: they may not be technical.
4. If the spec is missing, stop and recommend completing Phase 0 (guide §5).

## Default mode (initialize)

### Step 0 — Read the inputs
Find and read, in this order (whatever exists):
- Technical spec: `docs/superpowers/specs/*-design.md` (Superpowers) or `spec.md`
- `docs/product/PRD.md`, `docs/product/market-research.md`
- `docs/design/PRODUCT.md`, `docs/design/DESIGN.md`

If there are several candidate specs, ask which one to use.

### Step 1 — Profile → `docs/project-profile.md`
Answer: project type (web, API, conversational bot, CLI, library, data pipeline, mobile; can be mixed), external interfaces, risks (money, personal data, security, legal), non-deterministic AI behavior, technical unknowns, approximate size, and non-negotiable rules.

### Step 2 — Apply the decision matrix
Follow `references/decision-matrix.md` to decide:
- the verification strategy (which tooling goes in `verify/`);
- the size and order of the list (walking skeleton → spikes → design system → risks → value → instrumentation → polish);
- each feature's `mode` (`sdd`, `inline`, `ultracode`, `spike`, `human`);
- which non-negotiable rules become deterministic controls (hooks and deny rules) and which become judgment (`CLAUDE.md` + a verification feature).

### Step 3 — Generate the artifacts
Use the templates in `templates/` and adapt them to the spec's stack:

| File | Template | Notes |
|---|---|---|
| Minimal skeleton | — | build, lint and test green with an empty test |
| `init.sh` | `templates/init.sh` | installs dependencies, validates `.env`, starts the environment, smoke test |
| `feature_list.json` | `templates/feature_list.example.json` | schema in `references/feature-schema.md` |
| `claude-progress.txt` | `templates/claude-progress.txt` | first entry = this session |
| `CLAUDE.md` | `templates/CLAUDE.md` | short; includes the session routine |
| `.claude/settings.json` | `templates/settings.json` | Stop hook with typecheck/test; deny rules for secrets |
| `verify/` | — | per the matrix (simulator, e2e, evals…) |
| `docs/decisions/` | — | empty; spike ADRs go here |

If the project has an interface: add design-system features early (tokens and base components from `DESIGN.md`). For every feature with UI, add a visual-evidence step and an empty/error-state step.
If the PRD defines metrics: add instrumentation features.

### Step 4 — Human checkpoint
Show:
1. The profile, in 5–8 plain lines.
2. The feature list as a table: id, description, `mode`, `risk`, `depends_on`, `human_review`.
3. Questions that help the review: *Does each feature fit in one session? Do the steps test real behavior? Does the order make sense? Did subjective items end up as `human`?*

Wait for approval or changes. On approval: `git add` + commit `"chore: initialize project memory (greenfield initializer)"`.

## `--resync` mode (the spec changed)
1. Read the current spec, PRD and DESIGN, and `feature_list.json`.
2. Compute the diff:
   - **New** → add them with `status: "pending"` and `passes: false`.
   - **Obsolete** → `status: "deprecated"`. **Never delete them.**
   - **Affected** (their behavior changed) → `passes: false`, `status: "pending"` and updated steps, with a note in `history`.
   - **Unchanged** → don't touch them.
3. Show the diff in plain language and wait for approval. Then commit and add an entry to `claude-progress.txt`.

## What this skill does NOT do
- It doesn't write the spec. That's Phase 0: Superpowers `brainstorming`, `product-management`, Impeccable.
- It doesn't build features. That's Phase 2: `/greenfield:next`.
Files in this package (8)
  • SKILL.md 5 KB
  • references/decision-matrix.md 2 KB
  • references/feature-schema.md 1 KB
  • templates/CLAUDE.md 1 KB
  • templates/claude-progress.txt 1 KB
  • templates/feature_list.example.json 2 KB
  • templates/init.sh 1 KB
  • templates/settings.json 1 KB

Versions

The database has no other skills named initializer.

Database snapshot: 13.11.2026 · package version facc7b9fe0c8 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100skill-089
    Initialize long-running agent projects with environment setup, feature list generation, and progress tracking. Use when starting a new implementation project or setting up an existing codebase for Claude-assisted development. Creates init.sh (dev server script),…
  • 100memory-init
    Initialize the Engram memory system — creates configuration, memory files, and CLAUDE.md integration. Trigger when the user sets up persistent memory for a project, or when `.claude/memory-settings.json` is missing and a memory feature is requested.
  • 90Project Scaffolding
    This skill should be used when the user wants to set up Claude for a project, create a CLAUDE.md file, scaffold the .claude directory, initialize an AI workflow, configure a project for Claude, set up dev intelligence, initialize a new project, or do a new project setup.…
  • 100scaffold-init
    Initialize an AI-native development workspace with spec-driven workflow infrastructure. Use when starting a new project, setting up Claude Code on a codebase for the first time, or when the user says scaffold, initialize, set up AI workflow, onboard this repo, or similar. Do not…
  • 89nacl-init
    Initialize a new project with CLAUDE.md containing development rules, skill routing, bug fix protocol, and documentation discipline. Use when: new project, initialize project, setup project, create CLAUDE.md, or the user says "/nacl-init".
  • 89nacl-init
    Initialize a new project with CLAUDE.md containing development rules, skill routing, bug fix protocol, and documentation discipline. Use when: new project, initialize project, setup project, create CLAUDE.md, or the user says "/nacl-init".

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.