@dzhechkov/skills-reverse-engineering

Reverse Engineering Unicorn — composite skill for Claude Code that reverse-engineers any company into an actionable launch playbook (plus, in DEEP mode, a clickable CJM prototype).

goap-research-ed25519 — self-learning (optional, since this release). When @dzhechkov/harness-cli is on PATH, the bundled research skill recalls prior METHOD lessons at the start of an investigation and records new ones at four named moments. Without it the skill behaves exactly as before and says so once — it is detected, never required. Lessons go to a SEPARATE store (<project>/.health-brain/.dz) and never to the shared one; recall reads both, so engineering lessons transfer in and medical ones do not leave. A format check refuses identifier shapes (email, phone, record numbers) — it does NOT judge whether a lesson describes a method or a person, and says so: that judgement is the agent's, per the teach protocol. See skills/goap-research-ed25519/SKILL.md.

What It Does

A 6-module pipeline orchestrated end-to-end, with three depth modes (QUICK / DEEP / VERIFIED):

1. M1 Intelligence — company/product intel gathering
2. M2 Product & Customers — JTBD, segments (+ M2.5 CJM prototype in DEEP)
3. M3 Market & Competition — landscape, positioning
4. M4 Business & Finance — model, unit economics
5. M5 Growth Engine — acquisition/retention loops
6. M6 Playbook Synthesis — the actionable launch playbook

What M5 produces, since 0.2.0 / 1.7.0 / 1.6.0

M5 no longer ends as analysis. Three changes, in the order they matter:

• Growth type is two independent choices, not one. A go-to-market MOTION (content, performance, sales-led, partnership, self-serve) and a growth LOOP (none, product-led, community, badge/embed, one- or two-sided incentivised referral, network effect). The old single list forced one pick across both questions, which made the ordinary case — a sales-led company running a referral loop — unsayable. Choosing no loop is a real answer and skips the loop-only output rather than demanding an invented flywheel.
• A Growth Requirements Seed table turns the analysis into FR-GROWTH-nnn DRAFT obligations, each naming the block it came from and carrying that block's confidence verbatim. A seed is a draft: it does not establish that anything was built.
• A compliance checklist runs before any technique becomes a requirement. It cites the norm and where to look it up, and deliberately carries no amount, threshold or statute — those are jurisdiction-specific and go stale within a year. A no answer is recorded against the requirement and blocks its promotion. It asks the questions; it is not a legal opinion.

Triggers on "проанализируй компанию", "reverse engineer", "разбери бизнес-модель", "playbook запуска".

Why

A standalone, canonical home for the orchestrator that several toolkits (keysarium, p-replicator) used to vendor as drifting copies. Centralizing it per ADR-0001 gives one source of truth; consumers reference it by id instead of carrying divergent forks.

How to Use

npx @dzhechkov/skills-reverse-engineering init

Then in Claude Code: /reverse-engineering-unicorn Notion or "разбери бизнес-модель Revolut".

What's Bundled

| Component | Description | |-----------|-------------| | reverse-engineering-unicorn | Orchestrator skill (6 modules, QUICK/DEEP/VERIFIED) | | explore | Pre-flight clarification (tracked vendor) | | goap-research-ed25519 | Adaptive research M1–M5 (tracked vendor) | | problem-solver-enhanced | Game Theory / TRIZ M3–M5 (tracked vendor) | | /reverse-engineering-unicorn | Slash command |

The core pipeline is self-contained. A few DEEP / Post-M6 modes reference extra skills that are not bundled (install separately if you need them):

| Optional dep | Used in | Get it from | |--------------|---------|-------------| | frontend-design | M2.5 CJM prototype (DEEP) | @dzhechkov/keysarium / p-replicator | | brutal-honesty-review | M6 BS-gate (DEEP) | @dzhechkov/skills-qe | | presentation-storyteller | Post-M6 pitch deck | @dzhechkov/keysarium | | idea2prd-manual | Post-M6 PRD | @dzhechkov/skills-idea2prd | | md2pptx | Post-M6 .md→.pptx | external |

Canonicalization (ADR-0001)

reverse-engineering-unicorn is the canonical artifact of this pack. It previously existed as two drifting copies (keysarium ≠ p-replicator); resolved here by taking the p-replicator body (which correctly references goap-research-ed25519, not the non-existent bare goap-research) plus keysarium's trust_tier frontmatter, and rewriting /mnt/skills/{user,public}/<id> → .claude/skills/<id>. The explore/goap/problem-solver trio is a sources.json-tracked vendored copy (canonical: @dzhechkov/skills-analyst-manual). Re-sync drift with dz sync-upstream.

CLI Commands

init (default) · update · remove · list · doctor — options --force, --dry-run, --help, --version.

Also Available Via dz

dz init --select reverse-engineering-unicorn
dz info reverse-engineering-unicorn

License

MIT

Signature scope (this release)

The pack's .dz-manifest.json now covers exactly the files this package SHIPS, as reported by npm pack — not everything present in the author's working tree. Previously it signed files that files[] excludes (typically CHANGELOG.md), so every recipient's verifier reported listed in the manifest but absent and the pack read as TAMPERED. Re-signing at any earlier moment could not fix that: those files were never in the tarball.

Nothing about the shipped content changed in this release — only what the signature describes.