Source Command Audit WhitepapersCAUTION

The most comprehensive Claude Code guide: agentic workflows, hooks, skills, MCP servers, quizzes, and production-ready templates. 430K+ lines.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: "source-command-audit-whitepapers"
description: "Audit version freshness, FR/EN parity, and metadata quality of all whitepapers and recap cards"
---

# source-command-audit-whitepapers

Use this skill when the user asks to run the migrated source command `audit-whitepapers`.

## Command Template

# Audit Whitepapers & Recap Cards

Comprehensive freshness and quality audit for all whitepapers (FR + EN) and recap cards (FR + EN). Scores each document against the current guide version, checks FR/EN parity, and validates metadata consistency.

## Arguments

- `--fix` - Generate update suggestions and frontmatter patches (does not modify files)
- `--verbose` - Show all criteria including passing ones, not just failures
- `--cards-only` - Audit only recap cards (skip main whitepapers)
- `--wp-only` - Audit only main whitepapers (skip recap cards)

## Usage

```bash
/audit-whitepapers                  # Full audit, failures only
/audit-whitepapers --fix            # Audit + prioritized update suggestions
/audit-whitepapers --verbose        # Full details for all criteria
/audit-whitepapers --cards-only     # Recap cards only
/audit-whitepapers --wp-only        # Whitepapers only
```

---

## Phase 1: Discovery

**Objective**: Locate and inventory all auditable documents.

### Steps

1. **Read current guide version**: Read the `VERSION` file at the project root. This is the target version all documents should be synced against.

2. **Scan directories** for `.qmd` files:
   ```
   whitepapers/fr/         (main whitepapers FR)
   whitepapers/en/         (main whitepapers EN)
   whitepapers/recap-cards/fr/    (recap cards FR)
   whitepapers/recap-cards/en/    (recap cards EN)
   ```

3. **Classify each file**:
   - **Main whitepaper**: files matching `[0-9][0-9]-*.qmd` in `whitepapers/fr/` or `whitepapers/en/`
   - **Cheatsheet**: files matching `*cheatsheet*.qmd` (tracked separately, not scored)
   - **Custom whitepaper**: files matching `*-whitepaper.qmd` or `*-cheatsheet.qmd` from external partners (strangebee, purchasely, devwithai) -- excluded from scoring, listed separately
   - **Recap card**: all `.qmd` files in `recap-cards/fr/` or `recap-cards/en/`
   - **Bundle/include**: `fiches-recap-serie.qmd`, files in `fiches/` subdirectory -- excluded (they are composed documents)

4. **Parse YAML frontmatter** for each included file (read between first `---` and second `---`). Extract:
   - `title`, `subtitle`, `author`, `date`
   - `version` (guide version tracked in main WPs and recap cards)
   - `wp-version` (whitepaper own semver, main WPs only)
   - `guide-version` (explicit guide version field used in recap cards)
   - `card-number`, `category`, `difficulty` (recap cards only)
   - `series` (main WPs only)

5. **Display discovery summary**:
   ```
   Guide version: {VERSION}
   Found: {N} main whitepapers ({FR}/{EN} FR/EN), {N} recap cards ({FR}/{EN} FR/EN)
   Excluded: {N} custom files (strangebee, purchasely, devwithai), {N} bundle files
   ```

---

## Phase 2: Version Gap Analysis (40 points per file)

**Objective**: Score how current each document is relative to the guide.

### Semver comparison logic

Parse version strings as `MAJOR.MINOR.PATCH`. The staleness metric is `minor_gap = current_minor - file_minor` (using the same MAJOR). If MAJOR differs, treat as critical.

For **main whitepapers**, the guide version field is `version`.
For **recap cards**, the guide version field is `guide-version` (fall back to `version` if absent).

### Scoring table

| Criterion | Points | Detection |
|-----------|--------|-----------|
| Guide version current (0-1 minor behind) | 15 | `minor_gap` <= 1 |
| Guide version recent (2-3 minor behind) | 10 | `minor_gap` 2-3 (partial credit, replaces the 15) |
| Guide version old (4-10 minor behind) | 5 | `minor_gap` 4-10 (partial credit) |
| wp-version field present and valid semver | 5 | Field exists AND matches `\d+\.\d+\.\d+` (WP only) |
| wp-version consistent with FR/EN pair | 10 | Same `wp-version` value in the counterpart file (WP only) |
| guide-version consistent with FR/EN pair | 10 | Same `guide-version` (or `version`) in the counterpart (cards) |

**Freshness labels**:
- 0-1 minor behind: Fresh
- 2-3 minor behind: Stale
- 4-10 minor behind: Very Stale
- 11+ minor behind: Critical

Note: The 15, 10, 5 pts for guide version are mutually exclusive (award only the highest applicable tier).

---

## Phase 3: Content Staleness Check (20 points per file)

**Objective**: Detect whether the guide sections a whitepaper covers have changed since it was last updated.

### Whitepaper-to-guide mapping (embedded)

| WP# | FR filename prefix | EN filename prefix | Guide sections |
|-----|--------------------|--------------------|----------------|
| 00 | `00-introduction` | `00-series-introduction` | `guide/ultimate-guide.md` Ch.1-2 |
| 01 | `01-prompts` | `01-effective-prompts` | Ch.2 prompting sections |
| 02 | `02-personnalisation` | `02-customization` | Ch.3 Memory, Ch.4 Agents, Ch.5 Skills |
| 03 | `03-securite` | `03-security` | Ch.7 Hooks, security sections |
| 04 | `04-architecture` | `04-architecture` | Ch.2 internals, architecture sections |
| 05 | `05-equipe` | `05-team` | Ch.3 team config, Ch.9 CI/CD |
| 06 | `06-privacy` | `06-privacy` | Ch.2 data/privacy sections |
| 07 | `07-guide-reference` | `07-reference-guide` | Ch.10 Reference |
| 08 | `08-agent-teams` | `08-agent-teams` | Ch.9 agent teams |
| 09 | `09-apprendre` | `09-learning` | `guide/roles/learning-with-ai.md` |
| 10 | `10-budget` | `10-ai-budget` | Business/ROI content |

### Steps

1. For each main whitepaper, extract the `date` field. If the date is not in `YYYY-MM-DD` format, parse it (e.g., "2026-02-12" or "March 2026").
2. Run `git log --since="{date}" --oneline -- guide/ultimate-guide.md guide/roles/ guide/core/` to count commits that touched guide content since the whitepaper's date.
3. Score:

| Criterion | Points | Detection |
|-----------|--------|-----------|
| No gui

CAUTIONgrade B · trust 89/100 Install with care. The audit found things worth knowing before you trust its output.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (2)

whitepapers/recap-cards/en/_extensions

whitepapers/recap-cards/fr/_extensions

Gates applied: no_behavioural_pass.