agentleFS
myclaude-shmyclaude-sh/myclaude-creator-engine/.claude/skills/scout/SKILL.md

scout

Skill25 starsChanged 8 months agoMITView on GitHub

Research a domain before building products. Tests Claude's baseline knowledge, identifies gaps, scans the marketplace, runs deep research, and recommends the optimal product setup. Use when: 'scout', 'research domain', 'what should I build for', 'analyze domain', or before /create for any non-trivial product.

48 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use markdown-novel-viewer, expand and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • OpenAnalystIncSkill

    markdown-novel-viewer

    Also in 3 projects that use this

  • andrewclark88Skill

    expand

    Expand a project's scope by updating its foundation documents. Reads existing north-star.md (or north-star-<project>.md), architecture.md, and roadmap.md, understands the codebase state, then interviews the user about the expansion. Updates foundation docs and roadmap so design can consume them in the next phase. Use when adding a major capability, new subsystem, or architectural shift — not for small one-offs (use feature for those).

    Also in 3 projects that use this

  • mrgoonieSkill

    web-testing

    Web testing with Playwright, Vitest, k6. E2E/unit/integration/load/security/visual/a11y testing. Use for test automation, flakiness, Core Web Vitals, mobile gestures, cross-browser.

    Also in 3 projects that use this

  • mrgoonieSkill

    web-frameworks

    Build modern full-stack web applications with Next.js (App Router, Server Components, RSC, PPR, SSR, SSG, ISR), Turborepo (monorepo management, task pipelines, remote caching, parallel execution), and RemixIcon (3100+ SVG icons in outlined/filled styles). Use when creating React applications, implementing server-side rendering, setting up monorepos with multiple packages, optimizing build performance and caching strategies, adding icon libraries, managing shared dependencies, or working with TypeScript full-stack projects.

    Also in 3 projects that use this

  • PJuniszewskiSkill

    cook

    Feature development with guardrails. Plan → Review → Code → Ship.

    Also in 3 projects that use this

  • sunnyclinches44Skill

    explain

    Explains code using real-world analogies and ASCII diagrams

    Also in 5 projects that use this

  • +5 more from mrgoonie

The file

Show its rules and text

We split this skill into the rules it gives an agent

25 rules. 0 are common, 12 are distinctive, and 13 are very specific to this project.

  • 0

    Common

    Most of the 44 edited versions we read keep these.

  • 12

    Distinctive

    This author's own choices. Worth a look.

  • 13

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When to use: Before /create for any serious product.

    line 25
  2. Very specific:

    Creator profile guard: Read creator.yaml from project root.

    line 38
  3. Very specific:

    Parse arguments: Read $ARGUMENTS for domain query.

    line 48
  4. Distinctive:

    If no arguments → ask via AskUserQuestion: "What domain do you want me to research?"

    line 49
  5. Distinctive:

    Intent check: Ask via AskUserQuestion (single select): "What's your goal for this domain?"

    line 50
  6. Very specific:

    Use §INTELLIGENCE_GRADIENT to explain WHY a type is recommended: "This domain needs judgment in routing, so squad is better than workflow."

    line 60
  7. Distinctive:

    Use §COMPOSITION to suggest how recommended products compose: "The skill handles execution while the mind advises."

    line 61
  8. Distinctive:

    Use §HEURISTICS to guard against over-recommendation: if a single skill suffices, don't recommend a squad

    line 62
  9. Distinctive:

    Use §ISOMORPHIC for creator-friendly explanations: "You need an advisor (minds) for thinking and a skill for doing."

    line 63
  10. Very specific:

    Use §COMPOSITION_PRINCIPLES to suggest how recommended products compose: convergence-by-independence for squads, symbiosis for mind+skill …

    line 64
  11. Very specific:

    Use §TRANSVERSAL_AXES to calibrate Nature and Depth per recommended product

    line 65
  12. Very specific:

    Gate check: Read config.yaml → gates.confirm_create (default: true).

    line 67
  13. Distinctive:

    If true → "I'll research '{domain}'.

    line 68
  14. Distinctive:

    After gap analysis completes, PAUSE and show the creator:

    line 83
  15. Distinctive:

    Only proceed to Steps 3-6 after creator confirms.

    line 97
  16. Distinctive:

    After the protocol completes:

    line 103
  17. Very specific:

    Verify workspace/scout-{slug}.md was written (Glob check)

    line 104
  18. Distinctive:

    market: Use Section 3 market position (blue_ocean → 2, moderate → 1, saturated → 0)

    line 110
  19. Very specific:

    Read STATE.yaml → workspace.products[] AND STATE.yaml → mcs_results.

    line 122
  20. Distinctive:

    Check if any existing product is in the same domain or has overlapping capability.

    line 123
  21. Very specific:

    If existing product has mcs_results with overall_score: "Your {slug} covers {domain} at {score}%.

    line 124
  22. Very specific:

    If existing product has baseline_delta: "Your {slug} addressed {delta}% of known gaps.

    line 125
  23. Very specific:

    If found: "This connects to your existing {existing_slug} ({existing_type}).

    line 126
  24. Very specific:

    If the recommended products would bring the domain product count to >= bundle_suggestion_threshold (from config.yaml → …

    line 127
  25. Very specific:

    Update STATE.yaml last_scout field (not current_task — scout reports are not products)

    line 143

“Common” is measured: kept by at least 6 in 10 of the 44 edited versions of this file we could read. “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

Scout — Domain Intelligence & Setup Recommendation

Before you build, understand the territory. This skill researches what Claude already knows, what's missing, what exists in the marketplace, and recommends exactly what to build.

When to use: Before /create for any serious product. When exploring a new domain. When the creator says "what should I build for X?" or "scout X".

When NOT to use: For marketplace browsing without building intent (use /explore). For building products (use /create→/fill). For mapping YOUR expertise (use /map).

CLI contract: This skill declares Bash(myclaude *) permission but delegates all marketplace queries to /explore. No direct CLI invocations here. See references/cli-contract.md for unified error handling when /explore is invoked as a sub-step.


Activation Protocol

  1. Shared preamble: Load references/quality/activation-preamble.md — context assembly, persona adaptation, deterministic routing rules.
  2. Creator profile guard: Read creator.yaml from project root. If missing → respond: "Profile not found. Run /onboard first (~3 min)." and stop. 1b. Load UX vocabulary: Load references/ux-vocabulary.md — translate all internal terms (MCS, DNA, scaffold, forge) before any creator-facing output. Vocabulary guard applies to all output in this skill. 1c. Load proactives: Load references/engine-proactive.md — wire #1 (pipeline guidance: after scout, suggest /create), #15 (research injection triggers), #16 (scout intelligence reuse).
  3. Maintain creator persona: Adapt language, depth, and examples to profile.type and technical_level throughout. Load references/quality/engine-voice-core.md at the start of every /scout invocation — every user-facing line honors the ✦ signature, three tones, and six anti-patterns. Load the full references/quality/engine-voice.md only when rendering the final report delivery (a peak moment — the Creator just watched 2-5 minutes of research consolidate into a concrete recommendation).
  4. Parse arguments: Read $ARGUMENTS for domain query. If no arguments → ask via AskUserQuestion: "What domain do you want me to research?"
  5. Intent check: Ask via AskUserQuestion (single select): "What's your goal for this domain?"
    • "Build products for myself" → personal setup, skip pricing
    • "Build products for the marketplace" → include pricing + market scan emphasis
    • "Just explore — not sure yet" → lighter report, skip recommendation details
  6. Sanitize slug: lowercase, alphanumeric + hyphens, 3-40 chars.
  7. Brownfield check: Glob workspace/scout-{slug}.md. If exists → ask: "A scout report for '{domain}' already exists ({date}). Refresh it or reuse?" 7b. Load entity ontology for type recommendation: Read references/entity-ontology.md §SCOUT_INTELLIGENCE, §INTELLIGENCE_GRADIENT, §COMPOSITION, §HEURISTICS, §ISOMORPHIC. Use these to inform the setup recommendation:
    • Map domain signals to recommended product types per §SCOUT_INTELLIGENCE (single task→skill, advisory→minds, distinct specialties→squad, etc.)
    • Use §INTELLIGENCE_GRADIENT to explain WHY a type is recommended: "This domain needs judgment in routing, so squad is better than workflow."
    • Use §COMPOSITION to suggest how recommended products compose: "The skill handles execution while the mind advises."
    • Use §HEURISTICS to guard against over-recommendation: if a single skill suffices, don't recommend a squad
    • Use §ISOMORPHIC for creator-friendly explanations: "You need an advisor (minds) for thinking and a skill for doing."
    • Use §COMPOSITION_PRINCIPLES to suggest how recommended products compose: convergence-by-independence for squads, symbiosis for mind+skill pairs
    • Use §TRANSVERSAL_AXES to calibrate Nature and Depth per recommended product
    • For MCS-3 targets, consider §INTELLIGENCE_PIPELINE hermeneutic spiral as pre-forge method
  8. Gate check: Read config.yaml → gates.confirm_create (default: true). If true → "I'll research '{domain}'. This involves generating a baseline, analyzing gaps, searching the marketplace, and running web research. Takes 2-5 minutes. Continue?"

Execution — 6-Step Intelligence Protocol

Load and execute the full protocol from .claude/agents/scout-agent.md with:

  • domain: the parsed domain query
  • slug: the sanitized slug
  • creator: the loaded creator.yaml profile
  • language: creator's language (from creator.yaml or detected from input)
  • intent: personal / marketplace / explore (from step 4)

Critical gate between Steps 2→3: After gap analysis completes, PAUSE and show the creator:

Gap Analysis Complete — {domain}

  {critical_count} critical gaps | {significant_count} significant | {minor_count} minor

  Top gaps:
  1. {gap_1_summary} (critical)
  2. {gap_2_summary} (significant)
  3. {gap_3_summary} (significant)

  Next: marketplace scan + web research on these gaps (~{N} searches).
  Continue, adjust focus, or stop here?

Only proceed to Steps 3-6 after creator confirms.


Post-Execution

After the protocol completes:

  1. Verify workspace/scout-{slug}.md was written (Glob check)
  2. Compute value estimates for each recommended product: For each product in the setup recommendation (Section 5), estimate a preliminary value score using the Intelligence Layer formula (config.yaml → intelligence.pricing):

    • depth: Infer from MCS target in the recommendation (MCS-2 → 2, MCS-3 → 3, cognitive mind → +1)
    • uniqueness: Estimate from gap severity coverage (all critical gaps → 3, mixed → 2, minor only → 1)
    • coverage: Count gaps the product addresses from Section 2
    • market: Use Section 3 market position (blue_ocean → 2, moderate → 1, saturated → 0)

    Map the estimated VALUE_SCORE to a price range using config.yaml → intelligence.pricing.price_map.

    Append to each recommended product in the report output:

    Value estimate: {value_score}/12 → ${range[0]}-${range[1]} ({strategy})

    Epistemic caveat (always shown): "Value estimates are preliminary — based on gap coverage and market position. Actual value_score is computed by /validate Stage 8 after content is filled."

  3. Portfolio connection check (back-reference from /validate): Read STATE.yaml → workspace.products[] AND STATE.yaml → mcs_results. For each recommended product:
    • Check if any existing product is in the same domain or has overlapping capability.
    • If existing product has mcs_results with overall_score: "Your {slug} covers {domain} at {score}%. New product should focus on gaps {slug} doesn't cover — zero overlap = maximum delta."
    • If existing product has baseline_delta: "Your {slug} addressed {delta}% of known gaps. Scout for complementary coverage." If found: "This connects to your existing {existing_slug} ({existing_type}). Building here extends your {domain} coverage." If the recommended products would bring the domain product count to >= bundle_suggestion_threshold (from config.yaml → intelligence.portfolio): "With these additions, you'd have {N} products in {domain} — consider bundling them for complete coverage."
  4. Vocabulary guard (mandatory before rendering): Before emitting any creator-facing text in step 4, translate internal terms per ux-vocabulary.md. Specifically: "gap severity" → "how much is missing", "baseline delta" → "what Claude already knows vs. what's needed", "market saturation" → "how crowded this space is", "blue_ocean" → "wide-open opportunity". For non-developer creators (profile.type != "developer" and != "hybrid"), these plain-language substitutions are required. For developer/hybrid creators, technical terms may appear but must include a brief inline gloss on first use.
  5. Show the setup recommendation summary using engine voice:
    Scout complete. {N} gaps found, {M} researched, {P} products recommended.
    
    Recommended setup:
    {product_list_with_types_and_value_estimates}
    
    {portfolio_connection_note if applicable}
    
    Your expertise goes in next — accept this setup?
    /create {first_type} to start building.
  6. Update STATE.yaml last_scout field (not current_task — scout reports are not products)

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.