원클릭으로 Manus에서 모든 스킬 실행

mm-feature-rnd

스타6

포크0

업데이트2026년 6월 9일 23:24

Feature R&D via a dynamic workflow. Mine a codebase's roadmap, scaffolded/gated features, ADRs, and novel ideas for net-new feature candidates; generate full design specs; score them with a judge panel on value/effort/risk/differentiation; synthesize a ranked HTML feature catalog. Use when the user asks "what should we build next", "feature ideas", "R&D", "find net-new features", or wants a decision-ready catalog of buildable features (not bug fixes). Read-only — produces designs + a catalog, writes no feature code.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

wolverin0

wolverin0/memorymaster

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

mm-feature-rnd — Feature R&D Catalog (dynamic workflow embedded in a skill)

Runs a multi-agent dynamic workflow that discovers net-new feature candidates from real project evidence, designs each one, scores them with a judge panel, and renders a ranked HTML catalog you pick from. This is the harness that produced a 17-feature scored catalog for MemoryMaster (2026-06-01).

When to use

"what should we build next", "feature ideas", "R&D session", "roadmap mining"
You want grounded, buildable feature proposals (scored, with full specs) — NOT generic brainstorming, NOT bug fixes/refactors (those are out of scope).

How it works (phases)

Scout — parallel agents mine: roadmap/backlog files, scaffolded/env-gated half-built features, ADRs/architecture for under-exploited primitives, and novel differentiators vs competitors. Curate to a deduped slate.
Design — one full design spec per candidate (problem, design on the real stack, surface, effort S/M/L/XL, risk, differentiation, test plan, deps).
Judge — 2 judges per design (product-value + build-pragmatics lenses) score value/effort/risk/differentiation; deterministic in-code tally.
Synthesize — executive "build-first" list + themes.

Run it

Confirm scope — the repo + that this is features/improvements only (no bugs/cleanup). The embedded SCOUT_SLICES/CONTEXT/GROUPS are MemoryMaster-tuned; for another project edit the copy first.
Launch:
```
Workflow({ scriptPath: ".claude/skills/mm-feature-rnd/feature-catalog.workflow.js" })
```
Returns { stats, synthesis{executive_summary, build_first[], themes[]}, catalog[] }.
Render the HTML catalog from the .output file via a small Python script (severity/score-ranked cards, filterable by verdict) to C:\Users\pauol\artifacts\YYYY-MM-DD-<slug>.html, then open it. Don't pull the full result into context.
Present the top "build-first" features inline; let the USER pick which to prototype. Prototyping is a SEPARATE gated step (worktree builds + tests) — never auto-build features.

Adapting to another repo

Edit REPO, CONTEXT (project + competitors), and SCOUT_SLICES (which files to mine) at the top of the workflow .js. Keep the judge-panel + tally intact.

Guardrails

Read-only / design-only. Produces specs + a catalog; writes no feature code.
Features only. Drop anything that is a bug fix, test, refactor, or cleanup.
Trust scores, not verdict prose. The 2-judge median can skew labels pessimistic; rank by the numeric score + the synthesis build-first list.
Verify "already exists" claims before building later. Scout/design agents hallucinate both directions about what's already built — any follow-up build MUST grep/read the real code first (see ref mm-712f).

이 저장소의 다른 Skills

같은 저장소

gitnexus-cli

wolverin0/memorymaster

Use when the user needs to run GitNexus CLI commands like analyze/index a repo, check status, clean the index, generate a wiki, or list indexed repos. Examples: "Index this repo", "Reanalyze the codebase", "Generate a wiki"

2026-06-306

mm-audit

wolverin0/memorymaster

Deep adversarial code audit via a dynamic workflow. Fan-out finders across correctness/concurrency/security/perf/contract dimensions × module groups, dedup, then 3-skeptic adversarial refute-verify, completeness critic, and a ranked HTML findings report. Use when the user asks for a "deep audit", "adversarial audit", "find all the bugs", "security+correctness sweep", or wants an exhaustive review that goes beyond a single-context pass. Read-only — produces a report, does not change code.

2026-06-096

mm-mine-corrections

wolverin0/memorymaster

Mine your own correction history for recurring rules. A dynamic workflow harvests user-correction turns from the MemoryMaster verbatim corpus (857k+ rows), distills each to a prescriptive rule, clusters recurring themes, and proposes evidence-linked SKILLS / rule-claims (not CLAUDE.md bloat) — each backed by the source verbatim ids. Use when the user asks to "mine my corrections", "find the rules I keep repeating", "what do I keep telling Claude not to do", "turn my sessions into skills", or wants to harden behavior from real feedback. Read-only on the DB — proposes, does not auto-write rules.

2026-06-096

name	mm-feature-rnd
description	Feature R&D via a dynamic workflow. Mine a codebase's roadmap, scaffolded/gated features, ADRs, and novel ideas for net-new feature candidates; generate full design specs; score them with a judge panel on value/effort/risk/differentiation; synthesize a ranked HTML feature catalog. Use when the user asks "what should we build next", "feature ideas", "R&D", "find net-new features", or wants a decision-ready catalog of buildable features (not bug fixes). Read-only — produces designs + a catalog, writes no feature code.
version	1.0.0

mm-feature-rnd — Feature R&D Catalog (dynamic workflow embedded in a skill)

When to use

"what should we build next", "feature ideas", "R&D session", "roadmap mining"
You want grounded, buildable feature proposals (scored, with full specs) — NOT generic brainstorming, NOT bug fixes/refactors (those are out of scope).

How it works (phases)

Scout — parallel agents mine: roadmap/backlog files, scaffolded/env-gated half-built features, ADRs/architecture for under-exploited primitives, and novel differentiators vs competitors. Curate to a deduped slate.
Design — one full design spec per candidate (problem, design on the real stack, surface, effort S/M/L/XL, risk, differentiation, test plan, deps).
Judge — 2 judges per design (product-value + build-pragmatics lenses) score value/effort/risk/differentiation; deterministic in-code tally.
Synthesize — executive "build-first" list + themes.

Run it

Confirm scope — the repo + that this is features/improvements only (no bugs/cleanup). The embedded SCOUT_SLICES/CONTEXT/GROUPS are MemoryMaster-tuned; for another project edit the copy first.
Launch:
```
Workflow({ scriptPath: ".claude/skills/mm-feature-rnd/feature-catalog.workflow.js" })
```
Returns { stats, synthesis{executive_summary, build_first[], themes[]}, catalog[] }.
Render the HTML catalog from the .output file via a small Python script (severity/score-ranked cards, filterable by verdict) to C:\Users\pauol\artifacts\YYYY-MM-DD-<slug>.html, then open it. Don't pull the full result into context.
Present the top "build-first" features inline; let the USER pick which to prototype. Prototyping is a SEPARATE gated step (worktree builds + tests) — never auto-build features.

Adapting to another repo

Edit REPO, CONTEXT (project + competitors), and SCOUT_SLICES (which files to mine) at the top of the workflow .js. Keep the judge-panel + tally intact.

Guardrails

Read-only / design-only. Produces specs + a catalog; writes no feature code.
Features only. Drop anything that is a bug fix, test, refactor, or cleanup.
Trust scores, not verdict prose. The 2-judge median can skew labels pessimistic; rank by the numeric score + the synthesis build-first list.
Verify "already exists" claims before building later. Scout/design agents hallucinate both directions about what's already built — any follow-up build MUST grep/read the real code first (see ref mm-712f).