Talent matrix
Skills, tools, and talent across the lifecycle.
131 entries across the 12 phases. Each asset links to its canonical source. No paid placement.
Discovery & Evidence
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Startup validation | Garry Tan / GStack | gstack office-hours | Skill | Validate raw startup or product idea with forcing questions | grill-me; research synthesis |
| Plan stress test | Matt Pocock / Superpowers | grill-me | Skill | Pressure-test a raw idea or plan one question at a time | gstack office-hours; decision memo |
| Assumption mapping | Pawel Huryn / Product Compass | pm-product-discovery:identify-assumptions-new | Skill | Surface value, usability, viability, feasibility, and GTM assumptions | grill-me; pre-mortem |
| Customer interviewing | Pawel Huryn / Product Compass | pm-product-discovery:interview-script | Skill | Run structured interviews that produce comparable data | Dovetail template |
| Research synthesis | Dovetail | Dovetail | Tool | Tag, synthesise, and search qualitative research | Great Question; Sprig |
| Behavioural research | Sprig | Sprig | Tool | In-product micro-surveys and behavioural research | UserTesting AI; Maze |
| Research ops | Great Question | Great Question | Tool | Recruit, schedule, and synthesise customer research | Dovetail; UserTesting AI |
| Usability testing | Maze | Maze | Tool | Run unmoderated usability tests against prototypes and live products | UserTesting AI; Sprig |
| Moderated research | UserTesting | UserTesting AI | Tool | AI-assisted moderated and unmoderated user research | Maze; Great Question |
| Product analytics (AI) | Amplitude | Amplitude AI | Tool | Ask product-analytics questions in natural language | Pendo AI; Mixpanel AI |
Strategy & Scope
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Scope and ambition | Garry Tan / GStack | gstack plan-ceo-review | Skill | Challenge product plan scope, ambition, trade-offs | gstack office-hours; devil's advocate |
| Product strategy framing | Pawel Huryn / Product Compass | pm-product-strategy:product-strategy | Skill | Frame strategy as customer, problem, wedge, and proof | Wardley mapping; business model canvas |
| Outcome metric framing | Pawel Huryn / Product Compass | pm-marketing-growth:north-star-metric | Skill | Define one metric that captures customer value | OKR workshop; HEART |
| Decision capture | Matt Pocock / Superpowers | to-prd | Skill | Convert a scope decision into a PRD engineering can act on | pm-execution:create-prd; decision memo |
| PM drafting | Claire Vo / ChatPRD | ChatPRD | Tool | Draft PRDs and product artifacts from product prompts | Linear Agent; Productboard |
| Agentic planning | Linear | Linear Agent | Tool | Turn issue/project context into planning and execution workflows | Atlassian Rovo; Jira PD; Productboard |
| Agentic planning (enterprise) | Atlassian | Atlassian Rovo | Tool | Enterprise project and knowledge context for agent-assisted planning | Linear Agent; Jira PD; Confluence AI |
| Discovery board | Atlassian | Jira Product Discovery | Tool | Capture and prioritise product ideas alongside delivery work | Productboard; Aha! |
| Roadmap and prioritisation | Productboard | Productboard | Tool | Centralise feedback, ideas, and roadmap decisions | Aha!; Jira PD |
| Roadmap and prioritisation | Aha! | Aha! | Tool | Strategy-to-roadmap planning with goals and releases | Productboard; Jira PD |
| Product analytics | Amplitude | Amplitude | Tool | Behavioural analytics for outcome metrics | Mixpanel; Pendo |
Prototype & Design
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Prompt-to-prototype | Dan Cary / Anthropic | Claude Design | Tool / workflow | Generate branded interactive artifacts before a full PRD | Vercel v0; Figma AI; Lovable |
| Generative UI | Vercel | Vercel v0 | Tool | Generate UI prototypes and frontend starting points from prompts | Claude Design; Google Stitch; Figma AI |
| Prompt-to-UI | Google Stitch | Tool | Generate UI designs and frontend code from prompts; export to Figma | Claude Design; Vercel v0; Figma AI | |
| Design generation | Figma | Figma AI | Tool | Generate and transform design artifacts inside design workflow | Claude Design; Vercel v0; Framer AI |
| Prototype-to-app | Lovable | Lovable | Tool | Generate runnable app prototypes from prompts | Vercel v0; Bolt.new; Replit Agent |
| Prototype-to-app | StackBlitz | Bolt.new | Tool | Generate and iterate runnable web app prototypes | Lovable; Replit Agent; Vercel v0 |
| Prototype-to-app | Replit | Replit Agent | Tool | Generate runnable apps inside a browser-based dev environment | Lovable; Bolt.new; Vercel v0 |
| Design AI in product | Framer | Framer AI | Tool | Generate and edit production-ready marketing sites | Figma AI; Webflow AI |
| Accessibility checks | Deque | Axe DevTools | Tool | Automated WCAG checks during design and build | Lighthouse; Pa11y |
| Frontend design skill | Anthropic | frontend-design | Skill | Create distinctive, production-grade frontend interfaces with high design quality | ui-ux-pro-max; design-iterator |
| UI/UX design skill | Community | ui-ux-pro-max | Skill | UI/UX design intelligence across 50+ styles, palettes, and stacks | frontend-design; design-iterator |
| Design iteration skill | EveryInc | design-iterator | Skill | Iterate generated designs toward a working artifact | frontend-design |
| Journey mapping | Pawel Huryn / Product Compass | pm-market-research:customer-journey-map | Skill | Map the customer journey end-to-end | Service blueprinting |
Spec & System Design
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Domain-grounded plan review | Matt Pocock / Superpowers | grill-with-docs | Skill | Challenge plan against context docs, domain language, and ADRs | grill-me; ADR workflow |
| Plan writing | Jesse Vincent / Superpowers | writing-plans | Skill | Write implementation plans another engineer (or agent) can resume cold | ADR + RFC pair |
| PRD drafting | Pawel Huryn / Product Compass | pm-execution:create-prd | Skill | Draft a PRD with structure and acceptance criteria | to-prd; ChatPRD |
| User stories | Pawel Huryn / Product Compass | pm-execution:user-stories | Skill | Write user stories with explicit acceptance criteria | job stories |
| Spec-to-code | AWS | AWS Kiro | Tool | Turn prompts and requirements into specs, tasks, tests, implementation | Linear Agent; Atlassian Rovo; ChatPRD |
| Knowledge AI | Atlassian | Confluence AI | Tool | AI search and writing inside Confluence | Notion MCP; Mintlify |
| Knowledge MCP | Notion | Notion MCP | Tool / infrastructure | Expose Notion workspace to agents via MCP | Confluence AI; Glean |
| Diagrams as code | Eraser | Eraser | Tool | Generate architecture and flow diagrams from text | IcePanel; Structurizr |
| Architecture modelling | IcePanel | IcePanel | Tool | Collaborative C4 architecture modelling | Eraser; Structurizr |
Agentic Build
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Agentic coding | Anthropic | Claude Code | Tool | Agent-assisted coding in developer environment | Cursor; GitHub Copilot; Sourcegraph Amp |
| Agentic coding | OpenAI | OpenAI Codex | Tool | Coding agent for building, modifying, reviewing and shipping code | Claude Code; Cursor; Sourcegraph Amp |
| Coding workspace | Cursor | Cursor | Tool | Agentic IDE workflow for codebase navigation, edits, and review | Claude Code; GitHub Copilot; Windsurf |
| Agentic coding workspace | Windsurf | Windsurf | Tool | Agentic IDE workflow for codebase navigation, edits, and review | Cursor; Claude Code; OpenAI Codex |
| Autonomous SWE reference | Cognition | Devin | Tool | Benchmark for autonomous software engineering claims | OpenAI Codex; Claude Code; Windsurf |
| Repo-aware coding agent | Augment Code | Augment Agent | Tool | Agentic coding for large codebases and organisation-scale workflows | Cursor; Claude Code; OpenAI Codex |
| Inline code assistant | GitHub | GitHub Copilot | Tool | Inline code completion and chat across editors | Cursor; Tabnine; Continue |
| Repo-aware enterprise agent | Sourcegraph | Sourcegraph Amp | Tool | Repo-aware coding agent with enterprise controls | Augment Agent; Cursor |
| Terminal coding agent | Aider | Aider | Tool | Open-source terminal-based pair programmer | OpenHands; Cline |
| Open-source agent harness | OpenHands | OpenHands | Tool / harness | Open-source baseline for coding-agent experimentation | Aider; SWE-Agent |
| Compound engineering skills | EveryInc | ce:work | Skill | Frame and supervise agent work with acceptance criteria and context | Linear Agent issue prep |
| Subagent orchestration | Jesse Vincent / Superpowers | subagent-driven-development | Skill | Split work into parallel subtasks with explicit handoffs | Pair programming |
| TDD discipline | Jesse Vincent / Superpowers | test-driven-development | Skill | Require a failing test before any implementation | verification-before-completion |
| Context engineering | Murat Cankoylan | context-engineering-collection | Skill collection | Patterns for tightening prompts, codebase chunks, and tool boundaries | ce:work; manual prompt tuning |
Code Testing
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Browser test automation | Microsoft | Playwright | Tool | End-to-end browser testing with trace and video | Cypress; Selenium |
| Browser test automation | Cypress.io | Cypress | Tool | Developer-friendly end-to-end browser testing | Playwright; Selenium |
| Component testing | Testing Library | Testing Library | Tool | User-centric testing for UI components | Enzyme |
| Component workshop | Storybook | Storybook | Tool | Develop and test UI components in isolation | Ladle |
| Visual regression | Chromatic | Chromatic | Tool | Visual regression testing on top of Storybook | Applitools; Percy |
| Visual AI testing | Applitools | Applitools | Tool | Visual AI testing across browsers and viewports | Chromatic; Percy |
| Unit testing (Vite) | Vitest | Vitest | Tool | Fast unit-test runner for Vite/ESM projects | Jest |
| Unit testing (Node) | Jest | Jest | Tool | Unit-test runner for Node and frontend projects | Vitest |
| Unit testing (Python) | pytest | pytest | Tool | Python unit-test framework | unittest |
| Verification skill | Jesse Vincent / Superpowers | verification-before-completion | Skill | Run verification commands and confirm output before claiming work is complete | manual smoke tests |
| Debugging skill | Jesse Vincent / Superpowers | systematic-debugging | Skill | Hypothesis-driven debugging discipline | Bisect + logging |
Agent Evals
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| PM skills marketplace | Pawel Huryn / Product Compass | phuryn/pm-skills | Skill collection | Public PM skill repository across product workflows | GStack; Matt Pocock skills |
| Agent evals | OpenAI | OpenAI Evals | Tool / framework | Create repeatable evals for model and agent behaviour | LangSmith; Braintrust; Promptfoo |
| Agent evals and observability | LangChain | LangSmith | Tool | Evaluate, trace, and monitor model and agent behaviour | Braintrust; Langfuse; Arize Phoenix |
| Agent evals | Braintrust | Braintrust | Tool | Evaluate LLM and agent behaviour with traces and regression sets | Langfuse; Promptfoo; Arize Phoenix |
| Agent evals and tracing | Langfuse | Langfuse | Tool | Open-source LLM observability and evals | LangSmith; Helicone |
| Agent evals | Promptfoo | Promptfoo | Tool | Test prompts, models, and agent behaviour with repeatable suites | Braintrust; Langfuse; Arize Phoenix |
| Eval platform (enterprise) | Humanloop | Humanloop | Tool | Eval and prompt management for enterprise LLM apps | Braintrust; LangSmith |
| RAG evals | Ragas | Ragas | Tool / framework | Reference-free evaluation framework for RAG pipelines | DeepEval; OpenAI Evals |
| LLM eval framework | Confident AI | DeepEval | Tool / framework | Open-source eval framework with prebuilt metrics | Ragas; OpenAI Evals |
| LLM observability | Weights & Biases | W&B Weave | Tool | Trace, evaluate, and compare LLM applications | LangSmith; Langfuse |
| Agent observability | Arize | Arize Phoenix | Tool | Open-source LLM and agent observability | Langfuse; LangSmith |
| Coding-agent benchmark harness | Princeton / SWE-agent | SWE-Agent | Tool / harness | Benchmark issue-to-patch coding agents under reproducible harness | OpenHands; SWE-bench |
| A/B test analysis | Pawel Huryn / Product Compass | pm-data-analytics:ab-test-analysis | Skill | Analyse A/B experiments for product and prompt comparisons | Manual eval review |
| Cohort analysis | Pawel Huryn / Product Compass | pm-data-analytics:cohort-analysis | Skill | Segment users to expose aggregate eval blind spots | Trace sampling by segment |
Review, Release & Provenance
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| AI PR review | CodeRabbit | CodeRabbit | Tool | AI-assisted pull-request review with summaries and inline comments | GitHub Copilot Code Review; Graphite |
| AI PR review | GitHub | GitHub Copilot Code Review | Tool | Copilot-powered code review inside GitHub PRs | CodeRabbit; Graphite |
| Stacked PRs | Graphite | Graphite | Tool | Stacked PR workflow and AI review | GitHub Copilot Code Review |
| Supply-chain security | GitHub | GitHub Advanced Security | Tool | Secret scanning, code scanning, and dependency review | Snyk; Semgrep |
| SAST / SCA | Snyk | Snyk | Tool | Developer-first security scanning across code and dependencies | Semgrep; GitHub Advanced Security |
| SAST | Semgrep | Semgrep | Tool | Pattern-based static analysis for security and quality | Snyk; SonarQube |
| Build provenance | Linux Foundation | Sigstore | Tool / framework | Cryptographic signing and provenance for software artifacts | SLSA; in-toto |
| Feature flags / release | LaunchDarkly | LaunchDarkly | Tool | Feature flags, progressive delivery, and release gating | Flagsmith; Harness |
| Sandboxed execution | E2B | E2B | Tool / infrastructure | Sandboxed cloud runtime for agent-generated code | Runloop; Modal |
| Code review skill | EveryInc | ce:review | Skill | Review agent-authored work with an evidence bundle | requesting-code-review |
| Review request skill | Jesse Vincent / Superpowers | requesting-code-review | Skill | Frame a review with intent, risk, and what to verify | ce:review |
| Pre-mortem | Pawel Huryn / Product Compass | pm-execution:pre-mortem | Skill | Name failure modes before a high-risk release | Release readiness review |
| Commit discipline skill | Anthropic | git-commit-push-pr | Skill | Enforce small, atomic commits with explainable messages | Manual conventional commits |
| Branch closeout skill | Jesse Vincent / Superpowers | finishing-a-development-branch | Skill | Verify tests, evals, rollback plan, and evidence before merge | Release readiness review |
Knowledge & Context Layer
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Enterprise search | Glean | Glean | Tool | Permission-aware enterprise search and assistants | Dust.tt; Microsoft 365 Copilot |
| Workspace assistant | Dust | Dust.tt | Tool | Connect company data to assistants with permission awareness | Glean; Notion MCP |
| Workspace AI | Microsoft | Microsoft 365 Copilot | Tool | AI across Word, Outlook, Teams, and SharePoint | Gemini Enterprise; Amazon Q Business |
| Workspace AI (AWS) | AWS | Amazon Q Business | Tool | Generative assistant grounded in enterprise data on AWS | Microsoft 365 Copilot; Glean |
| Workspace AI (Google) | Google Cloud | Gemini Enterprise | Tool | Gemini-based enterprise assistant grounded in Workspace data | Microsoft 365 Copilot; Glean |
| RAG framework | LlamaIndex | LlamaIndex | Tool / framework | Data framework for connecting LLMs to private data | LangChain |
| Vector database | Pinecone | Pinecone | Tool / infrastructure | Managed vector database for retrieval | Weaviate; pgvector |
| Docs platform | Mintlify | Mintlify | Tool | Docs platform optimised for AI consumption and discovery | Read the Docs; GitBook |
| Compound knowledge skill | EveryInc | ce:compound | Skill | Compound knowledge into a citable graph; refresh on a schedule | Glean; Dust.tt |
Agent Operations
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| Observability (general) | Datadog | Datadog | Tool | APM, logs, and infrastructure monitoring | Grafana; New Relic |
| Observability (open) | Grafana Labs | Grafana | Tool | Dashboards, alerts, and observability on open standards | Datadog; Kibana |
| LLM observability | Helicone | Helicone | Tool | LLM logging, monitoring, and cost tracking | Langfuse; Portkey |
| LLM gateway | Portkey | Portkey | Tool / infrastructure | Routing, caching, and observability for LLM calls | Cloudflare AI Gateway; Helicone |
| LLM gateway (edge) | Cloudflare | Cloudflare AI Gateway | Tool / infrastructure | Edge gateway with caching, rate limits, and analytics | Portkey; Kong Agent Gateway |
Post-Release Feedback
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| AI support | Intercom | Intercom Fin | Tool | AI-first customer support agent with resolution metrics | Decagon; Sierra; Ada |
| Support platform | Zendesk | Zendesk | Tool | Help desk and ticketing with AI agents | Intercom Fin; Salesforce Agentforce |
| AI support agent | Decagon | Decagon | Tool | Custom AI support agents with QA and analytics | Intercom Fin; Sierra |
| AI support agent | Sierra | Sierra | Tool | Conversational AI agents for support and operations | Decagon; Intercom Fin |
| AI support automation | Forethought | Forethought | Tool | Triage, deflect, and assist agents inside support workflows | Ada; Sierra |
| AI support agent | Ada | Ada | Tool | AI customer service agents with brand and policy controls | Decagon; Sierra |
| CRM agents | Salesforce | Salesforce Agentforce | Tool | Agentic actions across the Salesforce platform | Microsoft Copilot Studio |
| Copilot platform | Microsoft | Microsoft Copilot Studio | Tool / infrastructure | Build copilots and agents with enterprise controls | Salesforce Agentforce |
| Sentiment analysis | Pawel Huryn / Product Compass | pm-market-research:sentiment-analysis | Skill | Extract sentiment and themes from support and feedback at scale | Manual QA scoring |
| User segmentation | Pawel Huryn / Product Compass | pm-market-research:user-segmentation | Skill | Segment users to expose blockers a single segment is hitting | Cohort analysis |
Agent Governance & Security
| Area | Person / source | Asset | Type | Jobs to be done | Primary alternatives |
|---|---|---|---|---|---|
| MCP standard | Anthropic | Anthropic MCP | Tool / standard | Standard for connecting agents to tools and data | Workato Enterprise MCP; Notion MCP |
| Governed enterprise actions | Workato | Workato Enterprise MCP | Tool / infrastructure | Govern agent tool access and enterprise action workflows | Tines; Kong Agent Gateway |
| Automation orchestration | Tines | Tines | Tool / infrastructure | No-code orchestration with audit trail; common agent action layer | Workato; n8n |
| Agent identity | Microsoft | Microsoft Entra Agent ID | Tool / infrastructure | Identity, permissions, and audit for AI agents | IAM; SPIFFE |
| Agent gateway | Kong | Kong Agent Gateway | Tool / infrastructure | Policy layer between agent and tools; deny by default | Cloudflare AI Gateway; Workato MCP |
| Model runtime guardrails | AWS | AWS Bedrock Guardrails | Tool | Apply safety and policy filters to model outputs | Lakera; Microsoft Copilot Studio |
| Prompt injection defence | Lakera | Lakera | Tool | Detect and block prompt-injection and policy violations | HiddenLayer; AWS Bedrock Guardrails |
| AI/ML security | HiddenLayer | HiddenLayer | Tool | Detect and respond to AI-targeted attacks | Lakera |
| Prompt injection and MCP security | Simon Willison / community | Prompt injection defense | Expert voice / practice | Understand and test indirect prompt injection and confused-deputy risk | Lakera; HiddenLayer |
| Security scan skill | OpenAI Codex | codex-security:security-scan | Skill | Run a security pass over agent-generated code | Snyk; Semgrep |
| Threat model skill | OpenAI Codex | codex-security:threat-model | Skill | Threat-model agent identity, scope, and tool boundaries | STRIDE workshop |