1Executive summary
The vault holds 78 Cloudworkz-built skills plus 15 imported external skills and 3 empty stubs. Only about half the Cloudworkz skills are reachable through Claude today, so the team cannot use most of them. Before migrating, this audit scores each skill against our own build standard, groups them into 14 functional categories, and flags duplicates and overlaps to clean up first.
35
Gold-standard (9–10) — migrate now
20
Solid / needs work (4–8.9) — polish then migrate
23
Thin or different-archetype (<4)
5
Retire (duplicates, superseded, empty stubs)
The headline for Werner
The discovery, content-compliance and curriculum-review families are genuinely gold-standard and ready to share. The cleanup is small and surgical: ~5 retirements (mostly duplicates and empty stubs) and a handful of merges. The one real build we recommend is a skills-orchestrator — a router so a teammate describes a task in plain language and the right skill (or the right sub-orchestrator) fires, instead of needing to know 70+ names.2How each skill was scored
Every SKILL.md was read and scored 0–10 against our own gold standard (the build-a-skill protocol and the 5-test). Points are earned for: a valid auto-discovery description (≤1024 chars), a Pass-0 context prime, four or more passes, the five closing sections (Hard rules / Performance / Failure modes / Related / Version), a complete sub-folder (notes + strategy + references), and a declared parent framework.
One important fairness note
The 0–10 score measures compliance with the Pass-protocol build shape. Four other archetypes legitimately don't use that shape — content-brain (MCP-strict, e.g.brain), engine (scheduled miners, e.g. daily-system-audit), command (imported slash-commands), and persona-agent (the EUK agents). A low score for these signals "different shape," not "low quality." Each skill below carries its archetype so the score reads fairly.3Quality distribution
Across the 78 Cloudworkz-built skills:
Accessibility gap: 24 of 78 Cloudworkz skills (31%) are reachable via Claude today. The legend ● = accessible now, ○ = not yet, appears beside every skill in the next section.
4Categories & per-skill scores
The derived scheme groups the library into functional families. Within each, skills are sorted best-first. ● = accessible via Claude today, ○ = not yet.
Discovery Engine
The AI-native discovery framework: ten ordered components plus three gap-fillers (API contract, user stories, UX flows), all driven by discover-orchestrator. The most mature, most internally-consistent family in the vault.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| scope-framer ○ | 10 | Pass-protocol | Discovery Component 1 skill. Given an org's product / service identity plus its sub-brand structure plus its strategic anchors, produces the seven-section Sprint 1 brief-and-scope output (Brief, What's In, What… | MIGRATE NOW |
| competitor-profile ○ | 10 | Pass-protocol | Discovery Component 3 skill. Given org positioning + a candidate competitor cluster list, produces Sprint 3 competitive-landscape output (cluster def + per-vendor profile rows + Cloudworkz-position-per-cluster … | MIGRATE NOW |
| persona-depth ○ | 10 | Pass-protocol | Discovery Component 4 skill. Given a role description plus the org's ICP plus the canonical product taxonomy, produces the Sprint 4 personas output (named persona + demographics + job-to-be-done + day-one scena… | MIGRATE NOW |
| grilling-agent ○ | 10 | Pass-protocol | Discovery Component 5 skill. Given a strategic claim (a Source of Advantage statement, a Play-to-Win cascade, a positioning thesis), pressure-tests it against named competitive clusters from Component 3 and aga… | MIGRATE NOW |
| ia-diagrammer ○ | 10 | Pass-protocol | Discovery Component 6 skill. Given an org's product taxonomy plus surface inventory (BOSS shell, CLOUDZ workspaces, Products, Modules, Settings, Marketplace, Billing), produces the Sprint 6 information architec… | MIGRATE NOW |
| architecture-diagrammer ○ | 10 | Pass-protocol | Discovery Component 7 skill. Given the org's substrate inventory plus the architect's lane-2 work plus product taxonomy, produces the Sprint 7 tech architecture output (layered architecture Mermaid + service in… | MIGRATE NOW |
| data-architecture ○ | 10 | Pass-protocol | Discovery Component 9 skill. Given an org's substrate plus the architect's schema work plus the canonical product taxonomy, produces the Sprint 9 data architecture output (entity list + Mermaid ERD + per-table … | MIGRATE NOW |
| roadmap-builder ○ | 10 | Pass-protocol | Discovery Component 10 skill. Given an org's OKRs plus the prior nine sprints (brief, problem statement, competitive, personas, play-to-win, IA, tech architecture, back-office, data architecture) plus stated se… | MIGRATE NOW |
| api-contract-specialist ● | 10 | Pass-protocol | Discovery API-contract skill. Given the schema (Component 9), the endpoint inventory, and the auth model, produces the API contract spec — endpoint groups, per-endpoint request/response shapes referencing real … | MIGRATE NOW |
| user-story-specialist ○ | 10 | Pass-protocol | Discovery user-story skill. Given the personas (Component 4), the IA (Component 6), the schema + API (Components 9/7), and the compliance ACU register, produces a complete INVEST user-story backlog with Given/W… | MIGRATE NOW |
| ux-flow-specialist ○ | 10 | Pass-protocol | Discovery UX-flow skill. Given the personas (Component 4), the IA (Component 6), and the feature catalogue, produces wireframe-level screen flows per persona journey — numbered step chips, screens, states (empt… | MIGRATE NOW |
| pain-pattern-aggregator ● | 9 | Pass-protocol | Pain-pattern aggregator. Scans a corpus (transcripts, Daily/ entries, research notes, inbox items, optionally Fireflies/Slack archives) for pain, friction, workarounds, expressed wishes. Clusters into named pai… | MIGRATE NOW |
| back-office-process-mapper ○ | 9 | Pass-protocol | Back-office process-mapper skill. Given a team description (sales-ops, marketing-ops, support, finance, legal, or equivalent) plus the org's tooling surface and product taxonomy, produces a six-section blueprin… | MIGRATE NOW |
| discover-orchestrator ● | 9 | Pass-protocol | Discovery orchestrator skill. Walks any AI-native discovery target (client engagement, internal product, investor narrative) through all 10 components of the AI-native discovery framework in sequence, firing th… | MIGRATE NOW |
Acid-Content & Compliance
The two-layer regulated-content substrate (verbatim-lock + expression-variants) and its programme orchestrator. Production-grade but light on closing sections.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| belief-gated-article-brief ○ | 6 | Pass-protocol | Belief-Gated Article Brief Production. Produces a single long-form article brief (T04) gated on a belief-stage transition: the piece must move a named persona from one belief to the next, draw only on approved … | REMEDIATE |
| event-template-content-brief ○ | 6 | Pass-protocol | Event-Template Content Brief. Instantiates a reactive standing brief (T06) per trigger type a tenant wants to be ready for, per the F12 reactive-trigger model: structure, format, compliance gate, turnaround, an… | REMEDIATE |
| content-programme-orchestration ○ | 6 | Pass-protocol | Content Programme Orchestration. Plans and assembles a full content programme (typically 20-30 pieces across 6-7 campaign sections) by breaking a belief journey into sections, dispatching a belief-gated brief (… | REMEDIATE |
| research-verify-lock ● | 5 | Pass-protocol | Research-Verify-Lock (RVL). Takes a candidate factual claim destined for regulated content, verifies it against an authoritative primary source, captures every mandatory qualifier, classifies it, and either loc… | REMEDIATE |
| expression-variant-generator ● | 5 | Pass-protocol | Expression Variant Generator. Takes a LOCKED ACU (a verified, qualified factual claim) and generates channel-specific, marketing-friendly expression variants that preserve every mandatory qualifier from the par… | REMEDIATE |
| prohibited-content-cataloguing ● | 5 | Pass-protocol | Prohibited Content Cataloguing. Systematically catalogues prohibited claims and phrases as T09 atoms with detection rules (literal/pattern/semantic) so future content auto-fails against them, feeding the PROHIB… | REMEDIATE |
| compliance-register-schema-design ○ | 5 | Pass-protocol | Compliance Register Schema Design. Stands up a compliance register (T02) for a new content programme in a regulated sector, the three-section LOCKED / expression-variants / PROHIBITED structure, sized and seede… | REMEDIATE |
Acid-Sales & Outreach
Outbound and call-side skills plus the five EUK-campaign agents. The EUK agents are persona/tone holders, not Pass-protocol skills — a different shape.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| live-call-coaching ● | 10 | Pass-protocol | Real-time call coaching skill. While [[Marie Ducher Clark]] (or another booker) is on a live call, surfaces real-time coaching prompts against the locked campaign script: where she dropped a section, where she'… | MIGRATE NOW |
| behavioural-email-sequence ● | 5 | Pass-protocol | Behavioural Email Sequence Constructor. Builds a 5-email sequence with behaviour-triggered branching (Day 4 split between opened-not-clicked and clicked-not-booked), a belief-progression arc, per-email word lim… | REMEDIATE |
| multi-channel-outreach-brief ● | 5 | Pass-protocol | Multi-Channel Outreach Brief Assembly. Assembles a 5-piece coordinated first-contact system (cold email universal + variant, voicemail, LinkedIn InMail, cold call x3 persona variants) deployed simultaneously at… | REMEDIATE |
| call-grading ○ | 3 | Pass-protocol | | Grade Aircall transcripts for any agent against the locked script for their campaign. Pulls transcripts from the agent's Drive folder, scores each call against the rubric, and emits four outputs: per-call s… | REMEDIATE |
| euk-campaign/euk-asset-coord ○ | 1 | Persona-agent | EUK Asset Coordination agent. Daily check on whether the materials Marie Ducher Clark and Roy Mendoza need for the next 24 hours of investor calls are ready, automating the manual daily coord loop. Reads the bo… | MIGRATE (archetype) |
| euk-campaign/euk-cold-tone ○ | 1 | Persona-agent | EUK Cold-list Tone agent. Holds the specific tone for the 30k cold list (people who have not heard from Enterprise UK in years), a sharper sub-positioning of the Positioning agent for the cold cohort. Produces … | MIGRATE (archetype) |
| euk-campaign/euk-compliance ○ | 1 | Persona-agent | EUK Compliance Firewall agent. Holds the "outside the FCA perimeter by design" line for the EUK campaign. Reviews every copy deliverable before send for advice-language drift, missing disclaimers, and missing s… | MIGRATE (archetype) |
| euk-campaign/euk-copywriting ○ | 1 | Persona-agent | EUK Copywriting agent. Writes the actual EUK-campaign deliverables (initial reach-out emails, follow-ups, newsletter pieces, website copy, ad copy, landing pages) drawing on the Positioning and Cold-list tone b… | MIGRATE (archetype) |
| euk-campaign/euk-positioning ○ | 1 | Persona-agent | EUK Positioning agent. Source-of-truth for Enterprise UK campaign tone across every deliverable, holding the non-threatening, non-hyperbolic, belief-led positioning and the sponsor-of-Unlock framing. Produces a… | MIGRATE (archetype) |
| investor-outreach-daily-analysis ○ | 1 | Engine | Daily analysis of ElevenLabs investor outreach calls — fetches yesterday's conversations, computes stats, generates concrete script-improvement suggestions, updates state.json + CSV exports. | MIGRATE (archetype) |
Unlock Curriculum Review
The eight-reviewer learning-centre panel plus its orchestrator. Boundaries between reviewers are explicitly written — low crossover risk, high quality.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| unlock-hnw-reviewer ○ | 9 | Pass-protocol | HNW persona reviewer for Unlock content and curriculum docs. Reads any Unlock-facing document from the perspective of the primary ICP (UK-resident HNW individual, £500k-£10M investable, exited founder or senior… | MIGRATE NOW |
| fca-guidance-checker ○ | 9 | Pass-protocol | FCA guidance-vs-advice boundary checker for Unlock content. Scans any Unlock-facing document for language that tips from guidance into advice, and flags it with severity, risk explanation, and a rewrite directi… | MIGRATE NOW |
| instructional-design-reviewer ○ | 8 | Pass-protocol | Instructional design reviewer for the Unlock learning centre curriculum. Applies adult learning principles to any Unlock curriculum document and returns a structured assessment: learning objectives, scaffolding… | POLISH |
| decumulation-specialist-reviewer ○ | 5 | Pass-protocol | Decumulation specialist reviewer for Unlock curriculum content. Reviews any content touching drawdown, pension sequencing, tax-wrapper ordering, longevity planning, or post-retirement strategy for accuracy, com… | REMEDIATE |
| behavioural-finance-reviewer ○ | 5 | Pass-protocol | Behavioural finance specialist reviewer for Unlock curriculum content. Reviews curriculum modules for alignment with how people actually make financial decisions. Identifies where cognitive biases will distort … | REMEDIATE |
| learning-centre-audit-orchestrator ○ | 5 | Pass-protocol | Orchestrator for the Unlock learning centre review team. Dispatches the full eight-skill review pipeline in correct sequence, collects findings, and produces a consolidated audit report with build-readiness ver… | REMEDIATE |
| financial-domain-reviewer ○ | 5 | Pass-protocol | Financial domain accuracy reviewer for Unlock curriculum content. Verifies every financial claim in a document against sourced reference material: investment principles, portfolio theory, UK tax mechanics, mark… | REMEDIATE |
| consumer-duty-reviewer ○ | 5 | Pass-protocol | FCA Consumer Duty and Targeted Support compliance reviewer for Unlock curriculum content. The FCA Targeted Support regime (live 6 April 2026, PS25/22) created a formal third category between generic guidance an… | REMEDIATE |
| bloom-taxonomy-reviewer ○ | 5 | Pass-protocol | Bloom's Taxonomy learning objective auditor for Unlock curriculum content. Audits every stated learning objective against the six cognitive levels of Bloom's Revised Taxonomy (Remember, Understand, Apply, Analy… | REMEDIATE |
| curriculum-coherence-auditor ○ | 5 | Pass-protocol | Whole-curriculum coherence auditor for the Unlock learning centre. Reviews the full module set as a single coherent system: vocabulary consistency, narrative arc integrity, cross-module contradictions, prerequi… | REMEDIATE |
Brand, Design & Render
The render and brand-asset pipeline. build-html is the spine; pdf-builder, iterate-copy-to-render and report-writing compose off it.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| build-html ● | 10 | Pass-protocol | HTML render builder. Produces a self-contained index.html in one of five named families (idea-presentation, render, canvas, script, report) inheriting style and structure from the family CSS block rather than r… | MIGRATE NOW |
| report-writing ○ | 10 | Pass-protocol | Report-writing skill. Produces a Cloudworkz market report (EUK annual edition, XLR8 sector report, special-edition brief) under the Enterprise UK publisher voice. Structure is operational, not fixed: the skill … | MIGRATE NOW |
| html-design-system-extractor ● | 9 | Pass-protocol | HTML design system extractor skill. Given a brand (Cloudworkz, Unlock, EUK, XLR8, Yachay, or new client) and a set of brand assets (live website URL, brochure / brand-board PDF, exemplar HTML, existing tokens.c… | MIGRATE NOW |
| iterate-copy-to-render ○ | 9 | Pass-protocol | Iterative copy-to-render pipeline. Walks a published surface's content through version-controlled markdown copy iterations, rendering each to HTML via build-html, then deploying. Generalises the EUK-Website / w… | MIGRATE NOW |
| pdf-builder ● | 9 | Pass-protocol | HTML-to-PDF conversion skill producing two professional-grade output flavours from any HTML source. Print-ready (CMYK-safe, bleed + crop marks, high-res, font-embedded, flattened for press) and read-ready (scre… | MIGRATE NOW |
| brand-foundation-builder ● | 4 | Pass-protocol | Brand foundation builder skill. Given a brand (Cloudworkz, Unlock, EUK, XLR8, Yachay) and a doc-type from the 6-doc Brand Box foundation stack (master-brand-document, play-to-win, comms-guidelines, brand-narrat… | REMEDIATE |
Stakeholder & Investor Comms
Targeted briefs for named stakeholders and investors, plus the Werner mini-app-brief generator (currently a stub).
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| investor-briefing-book ● | 10 | Pass-protocol | Investor briefing book builder. Produces a multi-investor pack with one master narrative plus per-investor deltas (highlights tailored to each investor's thesis, prior conversations, named concerns, ICP). Sibli… | MIGRATE NOW |
| build-stakeholder-handover ○ | 9 | Pass-protocol | Stakeholder-handover brief builder. Produces a targeted brief for one named stakeholder (Werner, Tom, Claudia, Fiorella, an investor, a contractor, a customer's named decision-maker), surfacing their priors plu… | MIGRATE NOW |
| mini-app-brief ● | 4 | Stub | STUB. Skill that generates Werner-sign-off-ready mini-app briefs in the format pinned by [[Intelligence/decisions/2026-05-29-mini-app-brief-format-V1]] - job + inputs + outputs + integrations + systems diagram.… | FINISH |
Quality & Review Gates
The pre-ship and rigour gates. Strong internal composition, but four of them overlap and need documented firing-boundaries.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| brief-gate ○ | 10 | Pass-protocol | Brief-gate skill. Given a structured task brief (typically the paste-block produced by the Cowork task-manager artefact, but any GOAL + ACCEPTANCE + SKILLS + DATA SOURCES brief works), runs a 7-criterion analys… | MIGRATE NOW |
| canvas-evaluate ○ | 10 | Pass-protocol | Decision-evaluation canvas skill. Walks a candidate decision through Frame, Solve, Cost, Decide passes (six-lens review inside Decide) and emits a verdict (PROCEED / DROP / WAIT) with named gaps. Generalises th… | MIGRATE NOW |
| thinking-quality ○ | 9 | Pass-protocol | General-purpose rigour check for any artefact (argument, plan, claim, brief, draft, decision, brainbox entry). Walks the artefact through eight anti-patterns (unfounded confidence, false binary, missing alterna… | MIGRATE NOW |
| stop-gate-review ○ | 9 | Pass-protocol | Pre-deploy stop-gate review skill. Standardised "before this ships" review with named gates. Given a build artefact (script, render, content piece, code merge, brief, deck) about to ship, walks it through eight… | MIGRATE NOW |
| section-rhythm-walk ○ | 9 | Pass-protocol | Section-rhythm walk skill. Walks a document section-by-section and surfaces where the pacing drags (long stretches without payoff), where it jumps (a leap the reader can't follow), where it loops (re-covering g… | MIGRATE NOW |
| grill-with-docs ● | 1 | Command | Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise. Use when user wants to stress-te… | CLARIFY |
Vault & System Ops
Vault health, routing, decision-recording and session handoff. Mixed archetypes (protocol, engine, command) sharing a vault-maintenance purpose.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| vault-manager ○ | 10 | Pass-protocol | Vault structure auditor and health manager. Audits the vault against the canonical Companies/Box/Acid schema, detects structural drift (missing folders, misplaced content, orphaned files, retired folders still … | MIGRATE NOW |
| record-decision ○ | 10 | Pass-protocol | Decision-artefact recorder. Writes a decision per the canonical schema (Resources/frameworks/decision-output-format-brief-V1), propagates to consumers (Tasks.md backlinks, affected project README updates), rout… | MIGRATE NOW |
| file-router ○ | 9 | Pass-protocol | File routing + QC skill. Ingests a file or batch, runs 5 QC lenses (content-type, project/tenant match, recency/version, live-vs-retired, sensitivity) to determine vault destination, reads file content for conf… | MIGRATE NOW |
| portal-audit ○ | 9 | Pass-protocol | Portal audit skill. Scores each of the 10 Cloudworkz OS Portal sections against a 4-point rubric (freshness, completeness, source alignment, audience fit), scouts the vault for content with no portal representa… | MIGRATE NOW |
| session-handoff ○ | 9 | Pass-protocol | Native session-handoff skill. Compacts the current session into a handoff document a fresh agent can pick up. Supersedes the Matt Pocock handoff mirror (Skills/handoff/SKILL.md) by replacing the mktemp default … | MIGRATE NOW |
| transcript-ingest ● | 7 | Pass-protocol | Transcript intake skill. Pass 0 primes from team roster + projects + recent decisions + canon so it never re-asks what the vault knows; Passes 1-6 extract entities, decisions (what each supersedes), framings (c… | POLISH |
| map-rebuild ● | 5 | Pass-protocol | Regenerates the Cloudworkz OS Vault Map from its manifest + the current vault, so the doc is a live representation of truth rather than a hand-maintained snapshot. Auto-refresh tabs (skills, templates, framewor… | REMEDIATE |
| daily-system-audit ○ | 0 | Engine | no description | MIGRATE (archetype) |
Ideation & Decision
The incubator (formerly BrainBox) ideation toolkit and its weekly miner, plus the lightweight decision-doc command.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| incubator ○ | 2 | Command | BrainBox ideation toolkit V2.3. Structures ideas through capture, four-prompt canvas (frame, solve, cost, decide), six-lens review panel inside Decide, and three verdicts (BUILD, SCRAP, WAIT). Use when Tom says… | CONVERT |
| incubator-candidates-weekly ○ | 1 | Engine | Sunday-evening miner ahead of the Monday Werner session. Scans ALL Claude sessions not yet reviewed since the last Incubator candidates ledger; surfaces a fresh ledger Tom reviews Sunday-Monday and picks for pr… | MIGRATE (archetype) |
| decision-doc ○ | 1 | Command | Structure a product/business decision into a vault-ready decision record — options, trade-offs, clear recommendation — filed to Intelligence/decisions/ in Cloudworkz OS house style. argument-hint: [decision to … | MERGE |
PM Command Utilities
Imported slash-command-style skills (argument-hint + allowed-tools). Useful but un-Cloudworkz-ified: no Pass-0 prime, no sub-folder, several redundant with gold-standard siblings.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| metrics-analyzer ○ | 1 | Command | Analyze product/business data from a CSV or pasted data — trends, anomalies, segments, actionable insights — and file the analysis as a vault-ready research record in Intelligence/research/. argument-hint: [pat… | CONVERT |
| okr-writer ○ | 1 | Command | Turn business goals into vault-ready OKRs — measurable Key Results, linked initiatives, alignment check, anti-goals — filed to Intelligence/considerations/ (or the relevant Project) in Cloudworkz OS house style… | CONVERT |
| prioritize ○ | 1 | Command | Score and rank a backlog with RICE/ICE into a vault-ready consideration record — ranked table, cut line, rationale — filed to Intelligence/considerations/ in Cloudworkz OS house style. argument-hint: [list of f… | CONVERT |
| technical-translator ○ | 1 | Command | Translate a technical document (API spec, architecture doc, RFC, eng proposal) into PM/product-impact language and file it as a vault-ready research record in Intelligence/research/, Cloudworkz OS house style. … | CONVERT |
| diagnose ● | 1 | Command | Disciplined diagnosis loop for hard bugs and performance regressions. Reproduce -> minimise -> hypothesise -> instrument -> fix -> regression-test. Use when user says "diagnose this" / "debug this", reports a b… | CONVERT |
| triage ● | 1 | Command | Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue wo… | CONVERT |
Meta & Skill-Building
The build protocol itself (build-a-skill) and the coding-discipline guardrail it invokes (karpathy-guidelines).
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| build-a-skill ● | 10 | Pass-protocol | Meta-skill that codifies the 8-step Cloudworkz skill build protocol. Produces a gold-standard SKILL.md + notes.md + strategy.md + references/ sub-folder from a brief (BrainBox entry, framework, or latent-capabi… | MIGRATE NOW |
| karpathy-guidelines ● | 1 | Command | Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success … | CONVERT |
Content Brain (MCP-strict)
The /brain strict-mode content retriever. A distinct archetype (MCP-only, no Pass structure) — and currently duplicated.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| brain ● | 1 | Content-brain | Cloudworkz Unlock content brain — strict mode. ACTIVATION RULE - ONLY invoke when the user's message begins with the literal prefix '/brain' (case-insensitive). Do NOT invoke based on topic match alone, even fo… | MIGRATE (archetype) |
| cloudworkz-brain ○ | 1 | Content-brain | Cloudworkz Unlock content brain — strict mode. ACTIVATION RULE - ONLY invoke when the user's message begins with the literal prefix '/brain' (case-insensitive). Do NOT invoke based on topic match alone, even fo… | RETIRE |
Handoff (legacy)
The superseded Matt Pocock handoff mirror. Replaced by session-handoff.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| handoff ● | 1 | Command | Compact the current conversation into a handoff document for another agent to pick up. argument-hint: What will the next session be used for? | RETIRE |
External / Imported
Third-party skills imported wholesale (Matt Pocock, Anthropic patterns). Not Cloudworkz-built and not held to the gold standard — catalogue and keep, do not migrate as house skills.
| Skill | Score | Archetype | What it does | Migration |
|---|---|---|---|---|
| external/brand-discover ○ | 2 | External | > | KEEP-EXTERNAL |
| external/brand-voice-enforcement ○ | 2 | External | > | KEEP-EXTERNAL |
| external/brainstorming ○ | 1 | External | You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation. | KEEP-EXTERNAL |
| external/caveman ○ | 1 | External | > | KEEP-EXTERNAL |
| external/deep-research ○ | 1 | External | Use when the user needs multi-source research with citation tracking, evidence persistence, and structured report generation. Triggers on "deep research", "comprehensive analysis", "research report", "compare X… | KEEP-EXTERNAL |
| external/dispatching-parallel-agents ○ | 1 | External | Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | KEEP-EXTERNAL |
| external/grill-me ○ | 1 | External | Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or… | KEEP-EXTERNAL |
| external/handoff ○ | 1 | External | Compact the current conversation into a handoff document for another agent to pick up. | KEEP-EXTERNAL |
| external/premortem ○ | 1 | External | Run a premortem on any plan, launch, product, hire, strategy, or decision. Assumes it already failed 6 months from now and works backward to find every reason why. Produces a revised plan with blind spots expos… | KEEP-EXTERNAL |
| external/subagent-driven-development ○ | 1 | External | Use when executing implementation plans with independent tasks in the current session | KEEP-EXTERNAL |
| external/to-prd ○ | 1 | External | Turn the current conversation context into a PRD and publish it to the project issue tracker. Use when user wants to create a PRD from the current context. | KEEP-EXTERNAL |
| external/verification-before-completion ○ | 1 | External | Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before as… | KEEP-EXTERNAL |
| external/writing-plans ○ | 1 | External | Use when you have a spec or requirements for a multi-step task, before touching code | KEEP-EXTERNAL |
| external/zoom-out ○ | 1 | External | Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture. | KEEP-EXTERNAL |
| external/boardroom ○ | 0 | External | | Boardroom convenes a multi-stakeholder council on a specific business decision the user is actively considering. Seven default roles (CEO, CFO, COO, CTO, General Counsel, CHRO, Board Chair acting as red tea… | KEEP-EXTERNAL |
5Redundancy & duplicates
Clean these up before migration so the team never sees two skills that do the same thing.
| Finding | Type | Detail | Recommended action |
|---|---|---|---|
| brain ≡ cloudworkz-brain | Duplicate | Same content-brain skill twice. brain carries full canon frontmatter + body and a migrated_from note; cloudworkz-brain is the thinner older copy. | Retire cloudworkz-brain; keep brain. |
| handoff → session-handoff | Superseded | session-handoff's own description states it supersedes the Matt-Pocock handoff mirror. A third copy sits in external/handoff. | Retire handoff; route all handoff intent to session-handoff. |
| decision-doc ↔ record-decision | Functional overlap | Both create a decision record in Intelligence/decisions/. record-decision is gold (10/10, propagates to Notion + backlinks); decision-doc is a thin command (1/10). | Merge decision-doc into record-decision. |
| brainbox / incubator | Rename residue | brainbox + brainbox-candidates-weekly folders are empty stubs; the live versions were renamed incubator + incubator-candidates-weekly. Yet build-html and canvas-evaluate still wikilink Skills/brainbox. | Delete the 3 empty stubs; fix the 2 stale wikilinks. |
| grilling trio | Crossover | Three interview/pressure-test skills: grilling-agent (discovery, strategic-claim pressure-test, gold), grill-with-docs (plan vs domain model, thin), external/grill-me (upstream import). | Keep grilling-agent; document grill-with-docs' niche or merge; tag grill-me external. |
| mini-app-brief stub | Drift | Vault SKILL.md is status: stub, but a fuller mini-app-brief is already exposed as an invokable skill — vault and invokable copies disagree. | Reconcile: finish the vault version as the single source. |
Plus 3 empty stub folders — brainbox, brainbox-candidates-weekly, brand-style-picker — contain no SKILL.md. Delete them.
6Crossover (overlapping scope)
These aren't duplicates — they're neighbours whose firing-boundaries should be written down so the right one triggers.
| Cluster | Skills | Overlap | Risk | Recommended action |
|---|---|---|---|---|
| Review gates | thinking-quality · stop-gate-review · brief-gate · canvas-evaluate | Four 'judge this artefact' gates. They already compose (declared edges between them), but a task could plausibly fire the wrong one. | Low–medium | Write a one-line firing-boundary into each description (brief = before work; thinking = mid-draft rigour; stop-gate = before ship; canvas = commit/no-commit on a decision). |
| FCA / compliance reviewers | fca-guidance-checker · consumer-duty-reviewer · financial-domain-reviewer | Three regulatory reviewers over Unlock content. Descriptions already distinguish them explicitly. | Low | No action — boundaries are documented. Keep as-is. |
| Competitive analysis | competitor-profile (discovery) · report-writing (market reports) | Both analyse competitors/market. Different outputs (discovery sprint vs published report) but shared research substrate. | Low–medium | Note the shared substrate; have report-writing consume competitor-profile output rather than re-research. |
| Vault health | vault-manager · file-router · portal-audit · daily-system-audit · map-rebuild | Five skills touch vault structure/health. Distinct jobs (audit / route / score-portal / daily-check / rebuild-map) but a 'tidy the vault' request is ambiguous. | Medium | Define a vault-ops entry point (see orchestrator) that routes to the right one. |
| Render pipeline | build-html · pdf-builder · iterate-copy-to-render · report-writing | Compose cleanly (all consume build-html). Crossover is intentional pipelining, not redundancy. | None | No action — healthy composition. |
7How skills interact
Skills already declare their relationships in frontmatter (related_skills, invokes). That gives us a real dependency graph. The most-referenced "hub" skills — the ones many others build on — are the highest-value migration priorities because everything downstream depends on them.
- build-html10 inbound links
- back-office-process-mapper8 inbound links
- transcript-ingest8 inbound links
- pain-pattern-aggregator7 inbound links
- instructional-design-reviewer6 inbound links
- scope-framer6 inbound links
- fca-guidance-checker6 inbound links
- grilling-agent5 inbound links
- stop-gate-review5 inbound links
- euk-campaign5 inbound links
Three orchestrators already exist
The library already has domain-level orchestrators that fire a whole family in sequence:| Orchestrator | Domain | What it does |
|---|---|---|
| discover-orchestrator | Discovery Engine | Fires all 10 components in sequence + manages handoffs |
| content-programme-orchestration | Acid-Content | Breaks a belief journey into sections, dispatches a belief-gated brief per piece |
| learning-centre-audit-orchestrator | Unlock Curriculum | Dispatches the 8-reviewer pipeline, consolidates findings |
8Proposed: a skills-orchestrator
You flagged we "might need a skills orchestrator." We agree — and the shape is clear from what's already here. Rather than a teammate memorising 70 skill names, a single router takes a plain-language task, classifies it to a category, and either fires the skill or hands off to the right domain orchestrator.
Layer 0 · Skills Router (new)
Plain-language intent → category → skill or sub-orchestrator. Reads the categorisation in this report + the live Skills Context Matrix. Refuses/clarifies when intent is ambiguous between two crossover skills.
Layer 1 · Domain orchestrators (exist)
discover-orchestrator · content-programme-orchestration · learning-centre-audit-orchestrator. Add a vault-ops entry point (routes the 5 vault-health skills) and a render entry point (build-html pipeline).
Layer 2 · Component skills (exist)
The 35 gold-standard skills and their solid siblings, fired individually or by an orchestrator.
Why build it
It is the single highest-leverage move for team adoption: it turns "which of the 70 skills do I need?" into "describe what you want." It also becomes the natural place to enforce the crossover firing-boundaries from §6, and it has a ready-made spec source inbuild-a-skill + the existing Skills Context Matrix.9Recommended migration sequence
- Clean up first (5 items). Retire
cloudworkz-brainandhandoff, delete the 3 empty stubs, fix the 2 stalebrainboxwikilinks, mergedecision-docintorecord-decision. - Migrate the 35 gold-standard skills now. They pass the build standard as-is — make them team-accessible immediately. Prioritise the hub skills from §7 since everything depends on them.
- Polish the 20 solid / needs-work skills. Most are one or two elements short (a sub-folder, or the closing sections). Batch-fix, then migrate.
- Decide on the command-style utilities. Either Cloudworkz-ify them (Pass-0 + sub-folder) or keep them deliberately as light commands with an explicit "command" tag.
- Catalogue but isolate the 15 external skills. Keep them, tag them external, don't hold them to the house standard or expose them as Cloudworkz skills.
- Write firing-boundaries for the crossover clusters (§6), then build the skills-orchestrator (§8) as the team's single entry point.
- Add a migration gate to
build-a-skillso every future skill ships team-accessible and categorised by default — no future re-accumulation of the backlog.