/detect-pack

Consolidate all detect outputs into unified index with risk heatmap and unknowns backlog.

---

What It Does

Reads all detect skill outputs and consolidates them into a unified knowledge index. Supports light mode (default, 1 summary file) and full mode (--full, 4+ detailed files with evidence index and unknowns backlog). Automatically detects whether each domain provided light-mode (summary) or full-mode (individual files) outputs.

---

Usage

/detect-pack [repo] [--full]

Argument	Required	Description
repo	No	Target repository (defaults to current)
--full	No	Run full consolidation (4+ output files with evidence index and unknowns). Default is light mode.

Light mode (default): Produces 1 summary file with risk heatmap, per-domain scores, and executive summaries.

Full mode (--full): Produces 4 files (single-platform) or 4 + per-platform packs (multi-platform), including evidence index, unknowns backlog, and cross-platform deduplication.

Mixed input handling: detect-pack automatically detects whether each domain produced light-mode (summary.md) or full-mode (individual files) outputs. Full-mode domains use per-file data; light-mode domains use summary frontmatter.

---

Output

Light Mode (default) — 1 file

File	Content
$JAAN_OUTPUTS_DIR/detect/summary.md	Risk heatmap, per-domain scores, executive summaries, input mode table

Full Mode (--full) — 4 files (single-platform)

File	Content
$JAAN_OUTPUTS_DIR/detect/pack/README.md	Knowledge index: metadata, domain summaries, overall score, links to all detect outputs
$JAAN_OUTPUTS_DIR/detect/pack/risk-heatmap.md	Risk heatmap table (domain x severity), top risks per domain
$JAAN_OUTPUTS_DIR/detect/pack/unknowns-backlog.md	Prioritized unknowns with "how to confirm" steps and scope boundaries
$JAAN_OUTPUTS_DIR/detect/pack/source-map.md	Evidence index: all E-IDs mapped to file locations

Full Mode — Multi-Platform Monorepo

File	Content
$JAAN_OUTPUTS_DIR/detect/pack/README-{platform}.md	Per-platform knowledge index (e.g., README-web.md, README-backend.md)
$JAAN_OUTPUTS_DIR/detect/pack/README.md	Merged pack: Consolidates all platforms into cross-platform risk heatmap and unified backlog
$JAAN_OUTPUTS_DIR/detect/pack/risk-heatmap.md	Cross-platform risk table with platform x domain matrix
$JAAN_OUTPUTS_DIR/detect/pack/unknowns-backlog.md	All platforms combined, grouped by platform then domain
$JAAN_OUTPUTS_DIR/detect/pack/source-map.md	All evidence IDs from all platforms

---

Multi-Platform Support

• Platform auto-discovery: Detects platforms by scanning for filename suffixes in detect outputs (e.g., stack-web.md, stack-backend.md)
• Per-platform packs: Creates separate pack for each platform at detect/pack/README-{platform}.md
• Merged pack algorithm: For multi-platform projects, creates additional merged pack that:
  • Aggregates findings across all platforms
  • Deduplicates cross-platform findings via related_evidence chains
  • Builds cross-platform risk heatmap (platform x domain table)
  • Combines unknowns backlog from all platforms
• Evidence ID parsing: Handles both formats:
  • Single-platform: E-DEV-001 → domain: DEV, platform: null, sequence: 001
  • Multi-platform: E-DEV-WEB-001 → domain: DEV, platform: WEB, sequence: 001
• Orchestration mode: If no outputs exist, asks if multi-platform, then displays platform-by-platform workflow guide

---

Orchestration (Step 0)

Before consolidation, checks which detect skills have run:

• No outputs found:
  • Asks: "Is this a multi-platform project? [y/n]"
  • If YES: Displays orchestration guide with platform-by-platform workflow
  • If NO: Lists all 5 detect skills and suggests running them first
• Partial outputs: Reports which domains are present/missing per platform, asks user to continue (results marked as partial)
• All outputs found: Proceeds directly to consolidation (per-platform + merged if multi-platform)

---

Seed Reconciliation (Step 8a)

After writing output files, detect-pack compares consolidated detection results against all project seed files and proposes updates:

• Reads: $JAAN_CONTEXT_DIR/tech.md, team.md, integrations.md, boundaries.md, tone-of-voice.template.md, localization.template.md
• Classifies changes: [UPDATE] (version drift), [ADD] (missing entry), [STALE] (not detected — user decides keep/remove)
• Presents diff-style summary per seed file with approval prompt: [y/all/n/pick]
• Applies approved changes preserving section anchors and user-added custom sections
• Suggests /learn-add for findings that don't map to any seed file
• Writes report: $JAAN_OUTPUTS_DIR/detect/seed-reconciliation.md

See seed-reconciliation-reference.md for mapping rules and preservation conventions.

---

ISO 25010 Compliance

detect-pack aggregates ISO/IEC 25010 quality characteristic mappings from all detect domains. The consolidated output includes:

• ISO 25010 coverage matrix — which characteristics are addressed by findings across domains
• Characteristic-level scores — aggregated score per ISO 25010 characteristic (Functional Suitability, Performance Efficiency, Compatibility, Usability, Reliability, Security, Maintainability, Portability)
• Gap analysis — identifies ISO 25010 characteristics with no coverage from any detect domain

Quality Gate Readiness

The pack output includes a Quality Gate Readiness section that evaluates whether the project meets configurable thresholds for shipping:

Gate	Default Threshold	Source
Overall score	>= 6.0	All domains
Critical findings	0	All domains
High findings	<= 3	All domains
ISO 25010 coverage	>= 6/8 characteristics	Mapped findings
Mutation score	>= 60% (if detected)	detect-dev testing.md

Gate thresholds are configurable via quality_gate in jaan-to/config/settings.yaml. The readiness verdict is PASS, WARN, or FAIL.

---

Key Points

• Enforces universal frontmatter: target.commit (must match git HEAD), target.platform, tool.rules_version, confidence_scheme, findings_summary, overall_score, lifecycle_phase (CycloneDX)
• Overall score formula: 10 - (critical×2.0 + high×1.0 + medium×0.4 + low×0.1) / max(total_findings, 1), clamped 0–10
• Risk heatmap:
  • Single-platform: domain × severity markdown table with per-domain scores
  • Multi-platform: platform × domain table showing findings per platform + totals row
• Evidence ID validation: all IDs must follow namespace convention, no duplicates, all resolve to file locations
  • Regex: ^E-([A-Z]+)-(([A-Z]+)-)?(\d{3}[a-z]?)$ (matches both single and multi-platform formats)
• Partial run handling: coverage % reported ("3/5 domains analyzed"), overall score labeled with "(partial)" suffix
• Unknowns backlog collects all findings with confidence ≤ Tentative + "how to confirm" steps
• Frontmatter validation failures become findings (severity: Medium, confidence: Confirmed)

---

Back to Detect Skills | Back to All Skills