Agent-Skills.md

Agent Skills: Roadmap Analyzer

Analyze project features against ICP (Ideal Customer Profile) needs to identify gaps and recommend roadmap priorities. Use this skill when asked to evaluate current product state, identify what should be built next, assess competitive positioning, or plan product roadmap based on target customer needs. Outputs gap analysis, prioritized backlog, and strategic themes. Discovers ICP and features from project documentation.

UncategorizedID: shipshitdev/library/roadmap-analyzer

Repository

shipshitdev/library
shipshitdev
151

Install this agent skill to your local

pnpm dlx add-skill https://github.com/shipshitdev/library/tree/HEAD/bundles/planning/skills/roadmap-analyzer

Skill Files

Browse the full folder contents for roadmap-analyzer.

Download Skill

Loading file tree…

bundles/planning/skills/roadmap-analyzer/SKILL.md

Skill Metadata

Name
roadmap-analyzer
Description
Turn a product's ICP into a revenue-ranked roadmap. Reads .agents/memory/icp.md (from the icp skill), inventories what already ships, finds the gaps that block landing/retaining/expanding the primary segment, and outputs a prioritized backlog plus strategic themes. Use when asked what to build next, how to prioritize the roadmap, what's blocking revenue, or to plan toward MRR. Hands off to roadmap-to-milestones.

Roadmap Analyzer

Rank what to build next by how much revenue it moves, grounded in the ICP. The output is a backlog a founder can defend to themselves: every item traces to landing, retaining, or expanding a paying segment.

Read-only on code. It produces analysis, not commits or issues — handoff to roadmap-to-milestones turns the backlog into tracked work.

Contract

Inputs:

  • .agents/memory/icp.md (from the icp skill) — the segments and their land/retain/expand levers. If absent, stop and recommend running icp first; a roadmap without an ICP ranks by taste, not revenue.
  • The codebase and product docs, to inventory what already ships.

Outputs:

  • Gap analysis: ICP need vs current state, severity-tagged.
  • Revenue-ranked backlog: P0/P1/P2 with a one-line revenue rationale each.
  • Strategic themes: 3–5 focus areas, sequenced.

Creates/Modifies:

  • None. Read-only.

Delegates To:

  • icp when no ICP doc exists yet.
  • roadmap-to-milestones to turn the ranked backlog into GitHub milestones + issues.
  • feature-intake / prd-writer to expand a single backlog item into a PRD.

Step 1 — Load the ICP

Read .agents/memory/icp.md. Pull out, for the primary segment: the acute pain, the buying trigger, the churn reasons, and the expansion path. These are the three revenue levers the roadmap serves:

  • Land — remove what blocks the primary segment from buying.
  • Retain — remove the named churn reason.
  • Expand — unlock the stated expansion path (seats, volume, tier).

If icp.md is missing, do not guess a customer. Say so and recommend icp.

Step 2 — Inventory what already ships

Discover current capabilities from the code and docs, not from memory. Search for the product's core actions, automation, collaboration, integration, and billing surfaces. For each capability record status (production / partial / missing) and quality (solid / rough / prototype), with a file:line anchor as evidence. Give credit for what exists — the gap is measured against reality, not a blank slate.

Step 3 — Gap analysis

Compare ICP needs against the inventory. One row per need:

| ICP need (segment) | Lever | Current state | Gap severity | |---|---|---|---| | <need> | Land/Retain/Expand | Missing/Partial/Solid | CRITICAL/HIGH/MEDIUM/LOW |

Severity is revenue-weighted: a Land or Retain gap for the primary segment is at least HIGH; a gap for a "not our ICP" segment is LOW no matter how large.

Step 4 — Rank by revenue, not by effort alone

Score each candidate:

  • Revenue impact (1–5) — does it Land a new sale, Retain against a churn reason, or Expand an account? Land/Retain for the primary segment score 4–5; polish for a secondary segment scores 1–2.
  • Urgency (1–5) — is a sale, renewal, or churn event blocked on it now?
  • Effort (1–5) — honest build size.

Score = (Revenue × 2 + Urgency) ÷ Effort. Higher = sooner.

Then apply the two rules that a naive score misses:

  • Finish over start. A feature that is partial and sits on a revenue lever is not a backlog item — it is a churn/lost-sale risk actively leaking money. Rank finishing it above starting anything new. A half-shipped core feature is a liability, not progress.
  • Sellable over impressive. Prefer the item that completes a story a customer will pay for over the one that adds a new half-story. Ten features at 80% sell worse than seven at 100%.

Bucket the ranked list P0 (build now), P1 (next), P2 (later), and pull out quick wins — high revenue impact, low effort — as the fastest MRR movers.

Step 5 — Strategic themes

Group the backlog into 3–5 themes, each named in the product's own language, each tied to a revenue lever. Sequence them: foundations that unblock the primary segment's Land first, then Retain, then Expand. For each theme give a one-line vision, the ICP segment it serves, and its success metric (ideally a revenue or retention number, not a vanity metric).

Output

# Roadmap Analysis — <product> (<date>)

## Executive summary
- <3–5 bullets: the revenue story and the single most important next move>

## Gap analysis
<table from Step 3>

## Revenue-ranked backlog
### P0 — build now
1. **<feature>** (score X.X) — Lever: <Land/Retain/Expand>. Why: <revenue rationale>.
### P1 — next
### P2 — later

## Quick wins
- <high-impact, low-effort items>

## Strategic themes
1. **<theme>** — serves <segment>, <lever>. Success: <metric>.

## Recommended next step
Run `roadmap-to-milestones` to sequence P0/P1 into milestones with due dates.

Anti-Patterns

  • Ranking without an ICP. Without icp.md, prioritization is preference. Run icp first.
  • Effort-first prioritization. Cheap-and-easy is not the same as revenue-moving; the score weights revenue ×2 for a reason.
  • Treating all segments equally. Optimize for the primary segment; a large gap for a non-ICP segment is still LOW.
  • Listing features without a revenue rationale. Every item names which lever it moves, or it does not belong on the backlog.
  • Rewarding new starts over finishing. Partial features on a revenue lever outrank anything new — they are leaking money now.