toolkit-scout

name	toolkit-scout
description	Concierge index for the Claude Code toolkit on this machine — router-first rule, by-intent map, workflow recipes, and rich family descriptions for the Mundi Princeps workflows. Curated head is hand-maintained; auto-generated inventory lives below the marker and is refreshed by ~/.claude/scripts/build-toolkit-scout.sh.

Toolkit Scout

⚠️ ROUTER-FIRST RULE. If you are not sure which tool to use for a task, run:

bash ~/.claude/scripts/route.sh "<task description>" --top=5

before scrolling this file or guessing from names. The router scores the full registry against your query and returns ranked matches with invocation syntax. Filter with --kind=skill|agent|command|mcp|script as needed.

This file is a concierge, not a phone book. Everything above the marker is hand-maintained: router-first rule → by-intent index → workflow recipes → family descriptions → when-stuck flowchart. The auto-generated inventory (counts, MCP list) is regenerated below the marker on every registry rebuild.

---

By intent — cross-cutting task → tool map

Map task types to the right tools. Use this when the router's keyword matcher doesn't surface what you need, or to check which families are relevant.

• Database / Supabase / Postgres → supabase (MCP), supabase-postgres-best-practices, postgres, postgres-optimization, ai-skills:postgres, migration-generator:create-migration, mcp__supabase__execute_sql
• PitchBook data → mcp__pitchbook__* (103 tools), /mundi:counterparty-enrich, /mundi:origination-run, mundi-orch-counterparty-enrich, mundi-orch-deal-origination, pitchbook-orchestrator agent
• CapIQ data → apps/capiq-mcp/ (81 tools), mundi-qmd-auth-refresh-hub for cookie refresh
• Deal origination (full pipeline) → /mundi:origination-run themes/<theme>.yaml → mundi-orch-deal-origination composes mundi-orch-counterparty-enrich + investment-banking:one-pager + private-equity:ic-memo
• Intent signals → /mundi:intent-signal-run → mundi-orch-intent-signal + mcp__pitchbook__pb_signal_* + mcp__perplexity__perplexity_research
• Counterparty enrichment → /mundi:counterparty-enrich → mundi-orch-counterparty-enrich + mcp__pitchbook__pb_entity_resolve + pb_signal_enrich
• Investor outreach → mundi-orch-investor-outreach + mcp__heyreach__* + mcp__pitchbook__pb_screen_investors + apps/lead-enricher/
• IC memos / CIMs / pitch decks / LBO / DCF → private-equity:ic-memo, investment-banking:cim, investment-banking:pitch-deck, financial-analysis:lbo-model, financial-analysis:dcf-model, mundi-orch-board-materials
• Cold outbound / LinkedIn → saraev-outbound-* family, saraev-cold-email-campaigns, cold-email, mcp__heyreach__*, linkedin-cli (Unipile-backed), link-drop-pipeline
• Design / landing pages / sites → brand, design-system, taste-skill, saraev-vibe-* family, saraev-cinematic-oneshot-site, /mundi:clone-site, /mundi:kw-subbrand, /mundi:investor-portal, /mundi:4-sites-one-shot
• Testing / Playwright / flaky tests → playwright (skill + mcp__playwright__*), testing:test-fix, bug-detective:debug, test-engineer agent, tdd, tdd-mastery
• CI / deploy → testing:test-fix, ci-debugger:analyze-ci-failure, devops:ci-pipeline, mundi-qmd-ssh-deploy-generalized, monorepo-tooling agent
• Claude Code config / hooks / agents → saraev-claude-code-advanced-hooks, git-guardrails-claude-code, update-config skill (for ~/.claude/settings.json)
• Video / YouTube → video-to-action, /mundi:video-to-spec, /mundi:youtube-first-check
• Multi-LLM debate / consensus / routing → /mundi:debate, /mundi:consensus, mundi-orch-multi-llm-consensus, mundi-orch-multi-llm-debate, mundi-orch-multi-llm-route, linkdrop-x-tom-doerr-multi-llm-council-deliberation
• PRD / product → prd-generator, write-a-prd, ariz-product-manager-toolkit, pm-execution:create-prd, pm-execution:write-prd, prd-to-plan, prd-to-issues
• Security audits → /mundi:security-audit, security:audit, security:hardening, mundi-qmd-secret-scan-precommit, mundi-qmd-fp-check-install, saraev-vibe-app-security-audit
• Board materials / investor decks → mundi-orch-board-materials, investment-banking:pitch-deck, financial-analysis:ppt-template-creator
• Auth refresh (PB / CapIQ / Supabase) → mundi-qmd-auth-refresh-hub, apps/pitchbook-mcp/scripts/refresh-pb-cookies.sh, apps/capiq-mcp/scripts/refresh-capiq-cookies.sh
• Session continuity / past-session lookup → mcp__qmd__deep_search, mcp__qmd__search, mcp__qmd__get, mcp__qmd__multi_get, ~/.claude/projects/-Users-mundiprinceps-Mundi-Princeps/memory/
• Document extraction / generation → document-skills:pdf, document-skills:docx, document-skills:pptx, document-skills:xlsx, document-skills:brand-guidelines, document-skills:doc-coauthoring
• Google Workspace → gws MCP (mcp__gws__* — Docs, Drive, Sheets, Calendar), ai-skills:google-drive, ai-skills:google-sheets, ai-skills:google-docs
• Microsoft 365 → ms365 MCP (mcp__ms365__* — Mail, Calendar, Teams, SharePoint)

---

Recipes — explicit workflow → tool-chain entries

Canonical workflows for Mundi's day-to-day. Each recipe names the primary tool, the orchestrator (if any), supporting tools, and safety constraints.

Recipe: Counterparty enrichment

• Primary: /mundi:counterparty-enrich "Bessemer Venture Partners" --kind=investor
• Orchestrator: mundi-orch-counterparty-enrich
• Supporting: mcp__pitchbook__pb_entity_resolve, pb_company / pb_investor / pb_lp / pb_advisor, pb_signal_enrich, mcp__perplexity__perplexity_research
• Safety: semaphore=2 on PitchBook. Never run parallel scrapers on same cookies (account ban risk — see memory feedback_no_parallel_scraping.md).

Recipe: Deal-origination pipeline from a theme

• Primary: /mundi:origination-run themes/<theme>.yaml
• Orchestrator: mundi-orch-deal-origination
• Composes: mundi-orch-counterparty-enrich, investment-banking:one-pager, private-equity:ic-memo
• Data: apps/deal-origination/themes/, mcp__pitchbook__pb_screen_companies

Recipe: Write an IC memo for a new deal

• Primary: private-equity:ic-memo
• Supporting models: financial-analysis:dcf-model, financial-analysis:lbo-model, financial-analysis:comps-analysis
• Data gathering: mcp__pitchbook__pb_ic_memo_data, pb_company_deep_dive, pb_management_assessment
• Format: document-skills:docx for final output

Recipe: Generate board materials from a data room

• Primary: mundi-orch-board-materials
• Data sources: Google Drive (mcp__gws__*), SharePoint (mcp__ms365__*)
• Deck construction: investment-banking:pitch-deck, financial-analysis:ppt-template-creator
• Extract: document-skills:pdf / docx / pptx

Recipe: Refresh PB / CapIQ cookies after 401

• Runbook: mundi-qmd-auth-refresh-hub
• Scripts: apps/pitchbook-mcp/scripts/refresh-pb-cookies.sh, apps/capiq-mcp/scripts/refresh-capiq-cookies.sh
• Critical: do NOT retry in a loop. See memory/feedback_pb_account_safety_protocol.md.

Recipe: Deploy an MCP to Achilles

• Runbook: mundi-qmd-ssh-deploy-generalized
• Template: apps/pitchbook-mcp/scripts/deploy-pb-counterparty-achilles.sh (canonical)
• Per-app ports: pitchbook=8766, capiq=8768, investor-outbound=8770

Recipe: Investor outreach campaign from thesis

• Primary: mundi-orch-investor-outreach
• Screener: mcp__pitchbook__pb_screen_investors
• Contacts: A-Leads + Apollo via apps/lead-enricher/ + mcp__heyreach__*
• Message gen: Gemini 3.1 Pro via mundi-orch-multi-llm-route
• Schedule: HeyReach campaign + Supabase CRM log
• Default: dry-run. --activate to go live.

Recipe: Intent-signal run on a theme

• Primary: /mundi:intent-signal-run signals/<theme>.yaml
• Orchestrator: mundi-orch-intent-signal
• Signals: mcp__pitchbook__pb_signal_advisor_hired, pb_signal_debt_maturing, pb_signal_fund_exits_due, pb_signal_management_changes, pb_signal_no_deal_in_years
• Grounding: mcp__perplexity__perplexity_research

Recipe: Multi-LLM decision deliberation

• Agreement-oriented: mundi-orch-multi-llm-consensus
• Adversarial: mundi-orch-multi-llm-debate
• Task→provider routing: mundi-orch-multi-llm-route
• External council: linkdrop-x-tom-doerr-multi-llm-council-deliberation
• Claude-only alternatives: /mundi:debate, /mundi:consensus

Recipe: LinkedIn profile / outreach automation

• Tools: linkedin-cli (Unipile-backed), mcp__heyreach__*, saraev-lead-scraper, saraev-outbound-* family
• Memory reference: linkedin-outbound-handoff.md (Unipile handoff pattern)
• Pipeline: Campaign YAML → Unipile SN search → filter open profiles → HeyReach list → campaign

Recipe: Postgres query against Kadenwood Supabase

• Primary: mcp__supabase__execute_sql
• Best practices: supabase-postgres-best-practices, postgres-optimization
• Migrations: migration-generator:create-migration, mcp__supabase__apply_migration

Recipe: Fix a flaky Playwright test

• Primary: playwright skill + mcp__playwright__*
• Debug: testing:test-fix, bug-detective:debug
• Agent: test-engineer

Recipe: Ship a code change to production

• Command: /ship (see ~/.claude/skills/ship-software/)
• CI gates: testing:test-coverage, review:code, review:security
• Deploy: mundi-qmd-ssh-deploy-generalized

Recipe: Audit CLAUDE.md / memory for staleness

• Command: /mundi:claude-md-audit
• Framework: Lehmann's 5 filters (relevance, freshness, model-version, redundancy, over-constraining)

Recipe: Write a PRD

• Primary: write-a-prd OR prd-generator
• Alternative: pm-execution:create-prd, pm-execution:write-prd
• Research first: ariz-product-manager-toolkit, ariz-product-discovery
• Convert to issues: prd-to-issues, prd-to-plan

Recipe: Generate a new landing page / site

• Primary: /mundi:clone-site, /mundi:kw-subbrand, /mundi:4-sites-one-shot
• Design: saraev-cinematic-oneshot-site, brand, taste-skill
• Design variants: design-system, saraev-3d-scroll-hero

Recipe: Secret scan / prevent commit leaks

• Install: mundi-qmd-secret-scan-precommit
• Manual: ~/.claude/scripts/phase3/check-secrets.sh
• Pre-rsync: ~/.claude/scripts/phase3/pre-rsync-achilles-grep.sh

Recipe: Lead enrichment pipeline

• Primary: apps/lead-enricher/
• Runbook: mundi-qmd-lead-enricher-supabase-writer (wires SupabaseWriter into waterfall)
• Providers: A-Leads → Apollo → PDL (waterfall cascade)

Recipe: Onboard a new MCP

• Audit: mundi-qmd-intent-tool-coverage
• Pattern: apps/pitchbook-mcp/ + apps/capiq-mcp/ as templates
• Deploy: mundi-qmd-ssh-deploy-generalized

Recipe: Past-session lookup / session continuity

• Primary: mcp__qmd__deep_search, mcp__qmd__search, mcp__qmd__get, mcp__qmd__multi_get
• Cross-session memory: ~/.claude/projects/-Users-mundiprinceps-Mundi-Princeps/memory/

Recipe: Ingest a URL into skills / KB / registry

• Agent: link-drop (invocable via Task(subagent_type="link-drop", prompt="<URL>"))
• Slash: /mundi:link-drop <URL>
• Skill: link-drop-pipeline

Recipe: Evolve an existing skill with new material

• Skill: autoresearch-evolve
• Note: additive only — never destructive rewrites

---

Skill families — what each one covers

Replaces raw slug-prefix counts with descriptions. Counts below are approximate and refresh automatically in the inventory table below the marker.

Family	Approx count	What's in it
saraev-cc-*	~182	Claude Code workflow tips from Saraev — hooks, n8n, terminal, voice, keyboard shortcuts, plan mode, worktrees
saraev-vibe-*	~83	"Vibe coding" patterns — SaaS scaffolds, one-shot sites, UI iteration, design feedback loops
saraev-biz-*	~74	Business patterns — agent design, outreach psychology, offer framing, retainer strategies
saraev-infra-*	~69	Infrastructure — Modal, Supabase, deployment, database schemas, security audits
saraev-outbound-*	~49	Cold outbound tactics — email, LinkedIn, forum autoresponder, casualize templates, AI tooling cost models
mundi-qmd-*	~33	Operational runbooks — auth refresh, secret scan, SSH deploy, env mgmt, async polling
mundi-orch-*	8	End-to-end orchestrators — counterparty, origination, intent signal, board materials, multi-LLM
linkdrop-x-*	~12	External community pointers (awesome-claude-code, council-of-high-intelligence, micrograd)
ariz-* (all subs)	~130	Consulting advisor personas — CPO, CTO, CMO, CFO, CISO, founder coach, market scan
geo-*	~11	GEO (Generative Engine Optimization) — citability, schemas, crawlers, llmstxt
gstack-*	~13	Product team orchestration — planning, shipping, freeze/unfreeze, worktrees
tob-*	~9	Trail of Bits security patterns — fuzzing, SARIF, semgrep, vulnerability scanners
hugging-*	~8	HuggingFace ecosystem — datasets, evaluation, model training, Gradio, paper publishing
hassid-claude-*	~6	Hassid's Claude 101 onboarding series
cloudflare-*	~5	Cloudflare agents, Workers, Durable Objects, Wrangler
better-auth-*	~5	better-auth library patterns (email/password, 2FA, organization)
expo-*	~7	Expo / React Native patterns
terraform-*	~6	Terraform modules, stacks, test, style guide
investment-banking:*	11	IB materials — CIM, teaser, pitch deck, merger model, buyer list, one-pager, process letter, datapack, strip-profile
financial-analysis:*	13	Financial models — 3-statement, DCF, LBO, comps, PPT templates, deck review, skill-creator
private-equity:*	9	PE workflow — DD checklist, IC memo, screen deal, value creation, portfolio, returns, add-on targets
document-skills:*	~17	Doc generation / manipulation — PDF, DOCX, PPTX, XLSX, brand guidelines, skill creator
apollo-pack:*	24	Apollo.io sales enablement — workflows, CI, observability, migration, rate limits, debug bundles
pm-*:* (8 packs)	~85	Product management — discovery, strategy, execution, market research, analytics, GTM, growth, toolkit
superpowers:*	14	Superpowers CLI — git worktrees, parallel agents, subagent-driven dev, TDD, code review
ai-skills:*	20	AI-first integrations — Gmail, Google Drive/Sheets/Calendar/Slides, Atlassian, Postgres, MSSQL, Imagen
mundi-* root	N/A	/mundi:* slash commands (21 workflows) — see docs/knowledge-base/wiki/00-workflows/INDEX.md

---

When stuck — flowchart

Can't find the right tool?

1. Run bash ~/.claude/scripts/route.sh "<task>" --top=5 — keyword + tie-break matcher
2. Miss? Check the "By intent" index above
3. Miss? Grep: grep -ri "<keyword>" ~/.claude/skills/*/SKILL.md | head
4. Miss? Check agents: ls ~/.claude/agents/ | grep -i <keyword>
5. Miss? Check slash commands: ls ~/.claude/commands/ | grep <keyword>
6. Still miss? mcp__qmd__deep_search for past sessions
7. Still miss? It probably doesn't exist yet — decide: write it, or use the closest approximation

---

Maintenance

• Add a new recipe: edit this file above the marker. build-toolkit-scout.sh preserves the curated head byte-for-byte on regen.
• Refresh counts: bash ~/.claude/scripts/build-toolkit-scout.sh (runs automatically after build-registry.sh).
• Marker invariant: exactly one <!-- AUTO-GENERATED BELOW --> marker must exist in this file. 0 or 2+ markers cause the regen to abort or fall back to the default template.

Auto-generated inventory

Registry built: 2026-04-22T00:52:44Z
Total items: 2015
Source: ~/.claude/registry.json
Regenerate: bash ~/.claude/scripts/build-toolkit-scout.sh

Counts by kind

Kind	Count
agent	171
command	344
mcp	16
script	204
skill	1280

Skill families (by slug prefix)

See curated head above for rich family descriptions. This table is raw counts only.

Family	Count
saraev-cc-*	182
saraev-vibe-*	83
saraev-biz-*	74
saraev-infra-*	69
saraev-outbound-*	49
mundi-qmd-*	33
apollo-pack:	24
ai-skills:	20
document-skills:	17
pm-execution:	15
superpowers:	14
pm-product-discovery:	13
linkdrop-x-*	12
pm-product-strategy:	12
geo-*	11
gstack-*	10
financial-analysis:	9
investment-banking:	9
private-equity:	9
tob-*	9
hugging-*	8
mundi-orch-*	8
ariz-marketing-*	7
context-mode:	7
expo-*	7
pm-market-research:	7
hassid-claude-*	6
hassid-skills-*	6
pm-go-to-market:	6
terraform-*	6
ariz-product-*	5
better-*	5
cloudflare-*	5
content-*	5
pm-marketing-growth:	5
saraev-claude-*	5
ariz-content-*	4
claude-mem:	4
hassid-cc-*	4
pm-toolkit:	4
(other standalone)	410

Agent families

Family	Count
gsd-*	12
geo-*	5
data-*	4
api-*	3
database-*	3
frontend-*	3
security-*	3
backend-*	2
code-*	2
compliance-*	2
documentation-*	2
error-*	2
fullstack-*	2
monorepo-*	2
performance-*	2
test-*	2
(other standalone)	120

Command namespaces

Namespace	Count
/gsd:*	32
/mundi:*	30
/(root):* (root slash-commands)	13
/git:*	7
/review:*	7
/architecture:*	6
/testing:*	6
/devops:*	5
/documentation:*	5
/refactoring:*	5
/security:*	5
/deploy-pilot:*	3
/doc-forge:*	3
/workflow:*	3
/a11y-audit:*	2
/accessibility-checker:*	2
/adr-writer:*	2
/ai-prompt-lab:*	2
/analytics-reporter:*	2
/android-developer:*	2
/api-architect:*	2
/api-benchmarker:*	2
/api-tester:*	2
/aws-helper:*	2
/azure-helper:*	2
/backend-architect:*	2
/bug-detective:*	2
/bundle-analyzer:*	2
/ci-debugger:*	2
/code-architect:*	2
/code-explainer:*	2
/code-guardian:*	2
/color-contrast:*	2
/commit-commands:*	2
/complexity-reducer:*	2
/compliance-checker:*	2
/content-creator:*	2
/contract-tester:*	2
/create-worktrees:*	2
/cron-scheduler:*	2

MCP servers

• clickup
• code-review-graph
• context7
• github
• gws
• heyreach
• ms365
• n8n
• pencil
• perplexity
• pitchbook
• playwright
• qmd
• supabase
• token-optimizer
• token-savior