Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
DeepSeek
Muse+22 moreClaude CodeCursorCodex CLIDeepSeekMuse
Dots
Manus
Grok
Grok Bot+18 agentsProduct Spec to Vertical Slices — MVP Build Plan & Walking Skeleton
Turns a PRD, product spec, or rough idea into a risk-weighted vertical-slice roadmap: what to build first, what to defer or cut, technical decisions to force early, and a walking skeleton that proves the architecture.
$15
Product Spec to Vertical Slices — MVP Build Plan & Walking Skeleton
Example session with this skill installed
Product: PalletLink — B2B wholesale order portal for specialty food distributors
Buyers browse ~800 SKUs, cart, and place orders for weekly delivery.
Hard lock: Monday 2:00pm local cutoff so warehouse can plan.
Warehouse needs pick lists grouped by delivery route.
Account managers need weekly order totals (no card payments in v1).
QuickBooks invoice export is desired later, not first release.
Non-goals: native mobile, driver GPS, multi-distributor marketplace.
Team: 2 full-stack + part-time designer · ~8 week first release.
Open: catch-weight pricing rule; late orders after cutoff?
Ask: vertical-slice build plan, walking skeleton, MVP cut-line, spikes, handoff seeds.
- Read your context and instructions
- Compiled the product spec to
Vertical Slice Implementation Roadmap
1. Spec Maturity Assessment
| Field | Value |
|---|---|
| Mode | Slice |
| Maturity level | 3 |
| Blocking gaps | Catch-weight pricing undecided; late-order policy undecided; auth only sketched |
| Ready to slice | yes (with assumptions) |
2. Capabilities (excerpt)
Catalog browse · Cart + place order · Cutoff lock · Route pick lists · AM weekly totals · QB export (later)
3. Architecture hypothesis
Web app + Postgres; server-enforced Monday 2pm cutoff in distributor timezone;
roles buyer / warehouse / AM; printable pick lists from locked orders;
catch-weight estimate with optional pick adjustment after ops decision.
Decision log (excerpt): catch-weight (estimate vs pick) · late orders (next window only) · auth (CSV-provisioned passwords for v1)
4–5. Vertical slices & walking skeleton
| ID | Slice | Effort | Confidence | MVP |
|---|---|---|---|---|
| VS1 | Walking skeleton: login → catalog from DB → cart shell | S | High | BUILD |
| VS2 | Cart + place order (pre-cutoff) | M | Medium | BUILD |
| VS3 | Monday 2pm hard lock (server time) | M | Medium | BUILD |
| VS4 | Route assignment + warehouse pick list | L | Medium | BUILD |
| VS5 | AM weekly totals dashboard | M | High | BUILD |
| VS6 | Catch-weight pick adjustment | M | Low | DEFER |
| VS7 | QuickBooks export | L | Medium | CUT (v1) |
Walking skeleton = VS1: real auth + catalog + deploy; stub cutoff/routes/picks.
Retires “can we ship a secured catalog app?” risk first.
6. Risk-weighted build order
- VS1 — retire deploy/auth/DB risk
- VS2 — core buyer value
- VS3 — launch-critical lock
- VS4 — warehouse can operate
- VS5 — commercial visibility
- VS6 — only after catch-weight spike
- VS7 — post-MVP finance
7. MVP cut-line
BUILD: VS1–VS5 · DEFER: catch-weight adjust, late-order override · CUT: QB, mobile, GPS, payments
8–9. Spikes & checkpoints
- Spike: cutoff TZ/DST fixtures → exit: boundary tests pass
- Spike: ops workshop on catch-weight → exit: written rule
- Checkpoint after VS3: any post-cutoff path in v1?
- Checkpoint after VS5: schedule QB or not?
10. Anti-patterns watched
Avoid warehouse-first horizontal build; avoid gold-plating QuickBooks/payments in v1.
11. Downstream handoff
Phases: P1 orders live (VS1–2) · P2 cutoff+picks (VS3–4) · P3 pilot visibility (VS5)
Backlog seeds: VS1–VS5 + spike tickets for TZ and catch-weight
12. JSON
Full Vertical Slice JSON: maturity, decision_log, slices[], build_order, risks,
mvp_verdict per slice, handoff.phase_seeds / backlog_seeds, labeled assumptions
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
- Product specs and PRDs often become horizontal task lists (all database, then all API, then all UI) so nothing is demoable until the end.
- Teams start coding without a walking skeleton, so integration and deploy risk explode mid-project.
- MVP scope balloons because there is no explicit build / defer / cut line for the first release.
- Technical choices (auth, pricing rules, third-party sync) stay implicit until they block delivery.
What it does
- Assesses spec maturity (levels 1–5) and chooses Slice, Shape, or Critique mode.
- Decomposes capabilities and user journeys with clear value statements.
- Proposes an architecture hypothesis and a technical decision log (options, trade-offs, decide-by).
- Creates end-to-end vertical slices with layers touched, integration points, effort, confidence, validation, and acceptance criteria.
- Defines the walking skeleton (what is real vs stubbed, which risk it retires).
- Builds a typed dependency graph and a risk-weighted build order with one-line rationale per position.
- Draws an MVP cut-line (BUILD / DEFER / CUT), spikes with exit criteria, decision checkpoints, and anti-pattern findings.
- Outputs phase seeds, backlog seeds, and machine-readable JSON for downstream planning.
Frameworks & tools
- Stack-agnostic product and feature planning (web, B2B, internal tools, marketplaces).
- Works from PRDs, feature briefs, Notion docs, or rough founder notes.
- Compatible with Claude Code, Cursor, Codex CLI, OpenClaw, and other SKILL.md agents—no SaaS roadmap tool required.
- Complements later sprint refinement and phase gating; this skill owns what/order/cut, not velocity math.
Why this beats prompting it yourself
- Enforces vertical slices and rejects pure layer tasks by design.
- Forces a walking skeleton first instead of big-bang delivery.
- Types dependencies (hard/soft × data/integration/knowledge/resource) so sequencing is defensible.
- Separates BUILD vs DEFER vs CUT so “nice-to-have” does not sink the launch.
- Requires spikes with exit criteria when confidence is low—no silent uncertainty.
- Anti-fabrication: maturity gaps and assumptions are explicit; estimates stay relative.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Product Management.
• Quality tracks how clear the input is; raw ideas need Shape mode first
• Effort is relative (S/M/L/XL), not calendar guarantees
• Not sprint capacity commitment or live project status reporting
• Does not write the full PRD, production code, or tickets
• Technical recommendations are hypotheses to validate
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 21 days ago
Permissions
This skill only reads the product description you provide and drafts a build plan. It does not implement features, open Jira/Linear issues, deploy services, or invent requirements as facts. Grant Read only; leave Shell, Write, Browser, Network, and Env off.