agentleFS
pinkroosteraipinkroosterai/PinkRoosterMcp/.claude/skills/pm-explore/SKILL.md

pm-explore

Skill2 starsChanged 8 months agoMITView on GitHub

Analyze the codebase from a product manager's perspective and suggest realistic, user-facing feature enhancements. Cross-references existing items to avoid duplicates. Creates selected suggestions as feature requests.

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 13 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈13

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every suggestion must be something a non-technical stakeholder would understand and care about.

    line 13
  2. Distinctive:

    If no arguments, use default limit of 5

    line 20
  3. Very specific:

    Call mcp__pinkrooster__get_project_status with projectPath set to the directory above

    line 25
  4. Distinctive:

    Use an Explore agent approach — read the codebase to understand what the application does from an end-user perspective, not a developer …

    line 44
  5. Distinctive:

    Use Glob, Grep, and Read to explore:

    line 55
  6. Distinctive:

    Do NOT suggest any of the following — they are developer concerns, not product features:

    line 81
  7. Distinctive:

    Before presenting, check each suggestion against the deduplication list from Step 2:

    line 94
  8. Distinctive:

    If a suggestion overlaps significantly with an existing FR, issue, or WP, discard it

    line 95
  9. Distinctive:

    If a suggestion partially overlaps, note the overlap and differentiate the scope

    line 96
  10. Very specific:

    After presenting the table, use the AskUserQuestion tool to let the user select:

    line 129
  11. Distinctive:

    If "none" was selected: "No feature requests created.

    line 166
  12. Distinctive:

    Every suggestion must be user-facing — something a non-technical stakeholder would understand

    line 171
  13. Distinctive:

    Never suggest refactors, performance optimizations, or developer tooling improvements

    line 172
  14. Very specific:

    Always cross-reference existing FRs/issues/WPs to avoid duplicates

    line 173
  15. Distinctive:

    Keep user stories concrete and specific, not vague platitudes

    line 175
  16. Distinctive:

    If the codebase is small or all obvious features are already tracked, it's OK to suggest fewer than the limit

    line 177

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Explore Feature Opportunities

Analyze the codebase like a product manager and suggest user-facing feature enhancements. Every suggestion must be something a non-technical stakeholder would understand and care about.

Step 0: Parse Arguments

Parse !arguments for flags:

  • --limit N: Maximum number of suggestions to generate (default: 5, max: 10)
  • If no arguments, use default limit of 5

Step 1: Resolve Project

  • Current directory: !pwd
  • Call mcp__pinkrooster__get_project_status with projectPath set to the directory above
  • Extract the projectId
  • Note the counts for context (how many issues, FRs, WPs exist)

Step 2: Load Existing Tracked Items

Load all existing items to avoid suggesting duplicates. Make these calls:

  1. mcp__pinkrooster__get_feature_requests with projectId and stateFilter: "Active"
  2. mcp__pinkrooster__get_feature_requests with projectId and stateFilter: "Inactive"
  3. mcp__pinkrooster__get_issue_overview with projectId and stateFilter: "Active"
  4. mcp__pinkrooster__get_issue_overview with projectId and stateFilter: "Inactive"
  5. mcp__pinkrooster__get_work_packages with projectId

Compile a deduplication list of all tracked item names and descriptions. You will cross-reference this list before presenting suggestions to ensure no overlap with existing work.

Step 3: Analyze the Codebase

Use an Explore agent approach — read the codebase to understand what the application does from an end-user perspective, not a developer perspective.

Focus on:

  • What does this application do? What problem does it solve?
  • Who are the users? What are their workflows?
  • What UI exists? What pages, forms, dashboards, and interactions are available?
  • What data entities exist and how do users interact with them?
  • What API endpoints exist and what capabilities do they expose?
  • What workflows or automation exist?

Use Glob, Grep, and Read to explore:

  • Dashboard pages and components (src/dashboard/src/pages/)
  • API controllers and routes (src/PinkRooster.Api/Controllers/)
  • MCP tools (src/PinkRooster.Mcp/Tools/)
  • Data entities (src/PinkRooster.Data/Entities/)
  • Shared DTOs and enums (src/PinkRooster.Shared/)
  • Skills and workflows (.claude/skills/)

Step 4: Generate Suggestions

Think like a product manager. For each suggestion, ask:

  • "Would a project manager or stakeholder request this?"
  • "Does this improve the user experience or unlock a new capability?"
  • "Is this distinct from existing tracked items?"

For each suggestion, produce:

  • Name: Concise, descriptive title (e.g., "Dashboard Export: CSV Download for Issue Lists")
  • Category: Feature (new capability), Enhancement (extends existing), or Improvement (makes existing better)
  • Priority: Critical, High, Medium, or Low
  • Business Value: 1-2 sentences explaining why a stakeholder would care
  • User Stories: 1-2 structured stories with role, goal, benefit

Generate up to --limit suggestions (default 5).

Explicit Exclusions

Do NOT suggest any of the following — they are developer concerns, not product features:

  • Refactors, code cleanup, or architectural changes
  • Performance optimizations or caching improvements
  • Library upgrades or dependency changes
  • Test coverage improvements
  • CI/CD pipeline changes
  • Infrastructure, deployment, or DevOps concerns
  • Code style, linting, or formatting improvements
  • Security hardening (unless it's a user-facing security feature like 2FA)
  • Documentation improvements (unless user-facing help/onboarding)

Step 5: Cross-Reference for Duplicates

Before presenting, check each suggestion against the deduplication list from Step 2:

  • If a suggestion overlaps significantly with an existing FR, issue, or WP, discard it
  • If a suggestion partially overlaps, note the overlap and differentiate the scope
  • Replace discarded suggestions with new ones to meet the --limit count

Step 6: Present Suggestions

## Feature Suggestions — {projectId}

**Analyzed**: {summary of what was explored}
**Existing items**: {frCount} FRs, {issueCount} issues, {wpCount} WPs

| # | Name | Category | Priority | Business Value |
|---|------|----------|----------|----------------|
| 1 | {name} | {category} | {priority} | {businessValue} |
| 2 | {name} | {category} | {priority} | {businessValue} |
| ... |

### Details

**1. {name}**
- Category: {category} | Priority: {priority}
- Business Value: {businessValue}
- User Stories:
  - As a {role}, I want {goal}, so that {benefit}
  - As a {role}, I want {goal}, so that {benefit}

**2. {name}**
...

---

After presenting the table, use the AskUserQuestion tool to let the user select:

  • Question: "Which suggestions would you like to create as feature requests?"
  • Header: "Create FRs"
  • multiSelect: true
  • Options: Build from suggestions (up to 4), e.g. [{label: "#1 {name}", description: "{category} | {priority}"}, {label: "#2 {name}", description: "{category} | {priority}"}, {label: "All", description: "Create all {N} suggestions as feature requests"}, {label: "None", description: "Skip creation — keep suggestions for reference only"}]

Step 7: Create Feature Requests

For each selected suggestion, call mcp__pinkrooster__create_or_update_feature_request with:

  • projectId
  • name: suggestion name
  • description: expanded description combining business value and context
  • category: mapped category
  • priority: mapped priority
  • businessValue: from suggestion
  • userStories: array of structured stories [{"Role": "...", "Goal": "...", "Benefit": "..."}]
  • requester: "pm-explore"

Collect all created FR IDs.

Step 8: Report Results

## Created {count} Feature Requests

| # | FR ID | Name | Priority | Status |
|---|-------|------|----------|--------|
| 1 | {frId} | {name} | {priority} | Proposed |
| ... |

### Next Steps
- Refine any FR: `/pm-refine-fr {frId}`
- Scaffold a work package: `/pm-scaffold {frId}`
- View project status: `/pm-status`
- Triage and prioritize: `/pm-triage`

If "none" was selected: "No feature requests created. Suggestions are available for reference above."

Constraints

  • Every suggestion must be user-facing — something a non-technical stakeholder would understand
  • Never suggest refactors, performance optimizations, or developer tooling improvements
  • Always cross-reference existing FRs/issues/WPs to avoid duplicates
  • Created FRs start as Proposed — no auto-state propagation
  • Keep user stories concrete and specific, not vague platitudes
  • The --limit flag controls suggestion count (default 5, max 10)
  • If the codebase is small or all obvious features are already tracked, it's OK to suggest fewer than the limit
  • Include specific next-step commands with FR IDs so the user can act immediately
  • For ambiguous suggestions (could be a bug or feature), note: "Unsure if bug or feature? Use /pm-plan <description> for guided classification"

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.