PackmindHub/packmind-onboard

packmind-onboard

Action skill. Provides complete automated onboarding for Packmind:

1. Creates or selects a package
2. Analyzes codebase for patterns
3. Generates draft Standards and Commands
4. Creates items via CLI

Automatic package creation when none exist, user selection when packages are available.

Guarantees

• Read-only analysis. Analysis phase does not modify any project files.
• Drafts before creation. All items are written as drafts first, allowing review before creation.
• Preserve existing. Never overwrite existing artifacts. If a slug already exists, create -2, -3, etc.
• Evidence required. Every reported insight must include file-path evidence (and line ranges when feasible).
• Focused output. Max 5 Standards and 5 Commands generated per run.
• Graceful failure. Partial failures don't lose successful work; failed drafts are preserved.
• User control. When packages exist, users confirm package selection before creation.

Definitions

• Pattern (non-linter): a convention a linter cannot reliably enforce (module boundaries, cross-domain communication, workflow parity, error semantics, etc).
• Evidence: path[:line-line] entries; omit line ranges only when the file isn't text-searchable.

---

Step 0 — Introduction

Print exactly:

I'll start the Packmind onboarding process. I'll create your first standards and commands and send them to your Packmind organization. This usually takes ~3 minutes.

---

Step 1 — Get Repository Name

Get the repository name for package naming:

basename "$(git rev-parse --show-toplevel)"

Remember this as the repository name for package creation in Step 2.

Also run packmind-cli whoami and extract the Host: value from the output. Remember this URL for the completion summary.

---

Step 2 — Package Handling

Handle package creation or selection.

Check existing packages

List available packages:

packmind-cli install --list

Parse the output to get package names.

No packages exist

Auto-create package using repository name:

packmind-cli packages create "${REPO_NAME}-standards"

Print:

No existing packages found — created a new one: ${REPO_NAME}-standards

One package exists

Ask via AskUserQuestion:

• "Add to {package-name}?"
• "Create new package instead"

Multiple packages exist

Ask via AskUserQuestion:

• List each existing package as an option
• Include "Create new package" option

If "Create new package" is selected

• Ask for package name (suggest ${REPO_NAME}-standards as default)
• Run: packmind-cli packages create <name>

Remember the selected/created package name for later reference.

---

Step 3 — Announce

Print exactly:

packmind-onboard: analyzing codebase (read-only)
Target package: [package-name]

---

Step 4 — Detect Existing Packmind and Agent Configuration

Before analyzing, detect and preserve any existing Packmind/agent configuration.

Glob (broad, future-proof)

Glob for markdown in these roots (recursive):

• .packmind/**/*.md
• .claude/**/*.md
• .agents/**/*.md
• **/skills/**/*.md
• **/rules/**/*.md

Classify

Classify found files into counts:

• standards: .packmind/standards/**/*.md
• commands: .packmind/commands/**/*.md
• other_docs: any markdown under .claude/, .agents/, or any skills/ or rules/ directory outside .packmind

If any exist, print exactly:

Existing Packmind/agent docs detected:

    Standards: [N]

    Commands: [M]

    Other docs: [P]

No overwrites. New files (if you Export) will be added next to the existing ones.

---

Step 5 — Detect Project Stack (Minimal, Evidence-Based)

Language markers (check presence)

• JS/TS: package.json, pnpm-lock.yaml, yarn.lock, tsconfig.json
• Python: pyproject.toml, requirements.txt, setup.py
• Go: go.mod
• Rust: Cargo.toml
• Ruby: Gemfile
• JVM: pom.xml, build.gradle, build.gradle.kts
• .NET: *.csproj, *.sln
• PHP: composer.json

Architecture markers (check directories)

• Hexagonal/DDD: src/application/, src/domain/, src/infra/
• Layered/MVC: src/controllers/, src/services/
• Monorepo: packages/, apps/

Print exactly:

Stack detected (heuristic):

    Languages: [..]

    Repo shape: [monorepo|single]

    Architecture markers: [..|none]

---

Step 6 — Run Analyses

Read each reference file for detailed search patterns, thresholds, and insight templates.

| Analysis | Reference File | Output focus | |----------|----------------|--------------| | File Template Consistency | references/file-template-consistency.md | Commands | | CI/Local Workflow Parity | references/ci-local-workflow-parity.md | Commands | | Role Taxonomy Drift | references/role-taxonomy-drift.md | Standards | | Test Data Construction | references/test-data-construction.md | Standards |

Output schema (internal; do not print as-is to user)

For every finding, keep an internal record:

INSIGHT:
title: ...
why_it_matters: ...
confidence: [high|medium|low]
evidence:
- path[:line-line]
where_it_doesnt_apply:
- path[:line-line]

---

Step 7 — Generate All Drafts

Generate all draft files in one batch, using the formats defined above.

Standard Draft Format

For each Standard insight, create a Markdown file at .packmind/standards/_drafts/<slug>.draft.md:

# Standard Name

What the standard covers and why.

## Scope

Where this standard applies (e.g., 'TypeScript files', 'React components').

## Rules

### Rule starting with action verb

Another rule can follow...

## Examples

### Good

```typescript
// Valid code example

Bad

// Invalid code example

### Command Draft Format

For each Command insight, create a Markdown file at `.packmind/commands/_drafts/<slug>.draft.md`:

```markdown
# Command Name

What the command does, why it's useful, and when it's relevant.

## When to Use

- Scenario when this command applies
- Another scenario...

## Checkpoints

- Question to validate before proceeding?

## Steps

### 1. Step Name

What this step does and how to implement it.

```typescript
// Optional code example

2. Another Step

Description of next step...

### Generation Rules

- Generate drafts **only from discovered insights** (no invention)
- Use evidence from analysis to populate rules/steps
- Cap output: max **5 Standards** + **5 Commands**
- Never overwrite existing files; append `-2`, `-3`, etc. if slug exists

---

## Step 8 — Present Summary & Confirm

Present the generated draft files and ask for confirmation:

============================================================ ANALYSIS COMPLETE

Target package: [package-name] Stack detected: [languages], [monorepo?], [architecture markers] Analyses run: [N] checks

DRAFTS CREATED:

Standards ([N]):

1. [Name] → .packmind/standards/_drafts/[slug].draft.md
2. ...

Commands ([M]):

1. [Name] → .packmind/commands/_drafts/[slug].draft.md
2. ...

Drafts are saved in .packmind/*/_drafts/ — you can review or edit them before creating.

Then ask via AskUserQuestion with three options:

- **Create all now** — Proceed with creating all standards and commands
- **Let me review drafts first** — Pause to allow editing, re-run skill when ready
- **Cancel** — Exit without creating anything

---

## Step 9 — Create Items

### If user selected "Create all now"

**IMPORTANT:** The CLI only accepts JSON playbook files, not markdown. Before calling the CLI, convert each `.draft.md` file to a `.json` file.

#### Standard JSON Schema

Convert the markdown draft to this JSON format:

```json
{
  "name": "Standard name (from # heading)",
  "description": "What the standard covers (from intro paragraph)",
  "scope": "Where it applies (from ## Scope section)",
  "rules": [
    {
      "content": "Rule starting with action verb (from ### Rule headings under ## Rules)",
      "examples": {
        "positive": "Valid code example (from ### Good section)",
        "negative": "Invalid code example (from ### Bad section)",
        "language": "TYPESCRIPT"
      }
    }
  ]
}

Command JSON Schema

Convert the markdown draft to this JSON format:

{
  "name": "Command name (from # heading)",
  "summary": "What it does and when (from intro paragraph)",
  "whenToUse": ["Scenario 1", "Scenario 2 (from ## When to Use bullets)"],
  "contextValidationCheckpoints": ["Question 1? (from ## Checkpoints bullets)"],
  "steps": [
    {
      "name": "Step name (from ### N. Step Name)",
      "description": "Step description (from step content)",
      "codeSnippet": "Optional code fence content"
    }
  ]
}

Conversion and Creation Process

For each standard draft:

1. Read the .draft.md file
2. Convert to JSON matching the schema above
3. Write the JSON to .packmind/standards/_drafts/<slug>.json
4. Run CLI command to create:

packmind-cli standards create .packmind/standards/_drafts/<slug>.json

5. If creation succeeded, add to package:

packmind-cli packages add --to <package-slug> --standard <slug>

6. Track result (success/failure)

For each command draft:

1. Read the .draft.md file
2. Convert to JSON matching the schema above
3. Write the JSON to .packmind/commands/_drafts/<slug>.json
4. Run CLI command to create:

packmind-cli commands create .packmind/commands/_drafts/<slug>.json

5. If creation succeeded, add to package:

packmind-cli packages add --to <package-slug> --command <slug>

6. Track result (success/failure)

Show progress:

Sending standards and commands to your Packmind organization...
✓ error-handling-pattern
✓ naming-conventions
✗ test-factory-patterns (error: duplicate name exists)
✓ run-full-test-suite

Done: 3 created, 1 failed

If user selected "Let me review drafts first"

Print:

Draft files ready for review at:
  - .packmind/standards/_drafts/
  - .packmind/commands/_drafts/

Edit them as needed, then re-run this skill to continue.

Exit the skill.

If user selected "Cancel"

Print:

Onboarding cancelled.
Draft files remain at .packmind/*/_drafts/ if you want to review them later.

Exit the skill.

---

Step 10 — Completion Summary

All items created successfully

============================================================
  ✅ ONBOARDING COMPLETE
============================================================

Package: [package-name]
Created: [N] standards, [M] commands

Your standards and commands have been created and deployed locally.

Next steps:
  - Reload your AI coding assistant to start using them
  - Visit [host from packmind-cli whoami] to manage your standards and commands
  - Run `packmind-cli install [package-slug]` in other repos to distribute them
============================================================

Clean up successful draft files after creation.

Partial success (some items failed)

============================================================
  ⚠️ ONBOARDING COMPLETED WITH ERRORS
============================================================

Package: [package-name]
Created: [N] standards, [M] commands
Failed: [X] items

Failed items:
  • [item-name]: [error message]

Failed drafts remain in .packmind/*/_drafts/ for review.
You can fix and re-run, or create manually with:
  packmind-cli standards create <file>
  packmind-cli packages add --to <package-slug> --standard <slug>
  packmind-cli commands create <file>
  packmind-cli packages add --to <package-slug> --command <slug>
============================================================

Keep failed draft files for user to fix and retry.

No patterns discovered

If analysis found no patterns:

============================================================
  ℹ️ NO PATTERNS DISCOVERED
============================================================

The analysis didn't find enough recurring patterns to generate standards or commands.

This can happen with smaller codebases or projects with very diverse coding styles.
You can try again later as the codebase grows, or create standards manually with:
  packmind-cli standards create <file>
============================================================

---

Edge Cases

Package creation fails

If packmind-cli packages create fails:

❌ Failed to create package: [error message]

Please check:
  - You are logged in: `packmind-cli login`
  - Your network connection is working
  - The package name is valid

Cannot proceed with onboarding until package is created.

Exit the skill. Do not proceed to analysis.

Not logged in

If CLI commands fail with authentication errors:

❌ Not logged in to Packmind

Please run:
  packmind-cli login

Then re-run this skill.

No packages available

If packmind-cli install --list returns no packages:

Auto-create a package using the repository name.

---

9.1 Deploy Locally (after successful creation)

Since the onboard skill is present, the user has configured an AI agent. Deploy the created artifacts locally using the package selected/created in Step 2:

packmind-cli install <package-slug>

This deploys to agent-specific folders:

| Agent | Standards | Commands | |-------|-----------|----------| | Claude | .claude/rules/packmind/standard-[slug].md | .claude/commands/packmind/[slug].md | | Cursor | .cursor/rules/packmind/standard-[slug].mdc | .cursor/commands/packmind/[slug].mdc | | Copilot | .github/instructions/packmind-standard-[slug].instructions.md | .github/prompts/packmind-[slug].prompt.md |

9.2 Cleanup and Summary

Delete the draft files, then print final summary:

============================================================
  PUBLISHED & DEPLOYED
============================================================

Standards and commands have been sent to your Packmind organization
and deployed to your AI coding assistant's configuration files.

Standards: [N]
  - [Name] (slug: [slug])
    → .packmind/standards/[slug].md
    → [agent-specific path]

Commands: [M]
  - [Name] (slug: [slug])
    → .packmind/commands/[slug].md
    → [agent-specific path]

Draft files cleaned up.
============================================================

If user declines (N):

Print:

Draft files ready for review at:
  - .packmind/standards/_drafts/
  - .packmind/commands/_drafts/

Edit them as needed, then re-run this skill to create them.