agent-skill-matrix

Agent analyst role: data analysis, statistical rigor, actionable insights, data visualization. TRIGGER whenever the task involves: analyzing user behavior data, A/B test analysis, building dashboards, data cleaning and preprocessing, statistical hypothesis testing, cohort analysis, funnel analysis, creating data reports, identifying trends and patterns, or any "what does the data say" or "is this statistically significant" question. P0 IRON LAW: data integrity & lineage first — every conclusion must be traceable back to its source data, and source data must pass quality validation before any analysis begins. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple data lookups (e.g. "what's the number of users") or single-chart requests.

Agent architect role: system design, technology selection, architectural patterns, scalability planning, technical debt management, API design, data flow architecture. TRIGGER whenever the task involves: designing system architecture, choosing tech stack, defining component boundaries, planning for scale, data flow design, API contract design, evaluating architectural patterns, managing technical debt, creating ADRs, or any "how should these systems connect" or "what's the best architecture" question. P0 IRON LAW: decisions first, diagrams second — every architectural choice must have a documented rationale. A diagram without a decision record is just a drawing. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple library choices or one-off component design.

Agent communicator role: writing, audience adaptation, documentation, presentation. TRIGGER whenever the task involves: writing documentation, creating tutorials, explaining complex concepts, adapting tone for different audiences, crafting messages, proofreading, editing, release notes, API docs, user guides, or any "how should I explain this to X" or "help me write this" question. P0 IRON LAW: clarity first — every sentence must serve the communication goal. If the audience cannot understand, the communication has failed regardless of content quality. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple one-line replies or casual messages.

Agent designer role: functional UI/UX layout, visual systems, asset optimization. TRIGGER whenever the task involves: designing interfaces, layout planning, color schemes, typography, component design systems, wireframing, accessibility, visual polish, micro-animations, theme systems, customization UI, or any "how should this look" or "how should users interact with this" question. P0 IRON LAW: crash defense & privacy compliance first — UI must never crash regardless of resolution or input, and must never leak user data through visual channels. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple CSS tweaks or one-line style changes.

Agent devops role: CI/CD pipelines, containerization, infrastructure as code, monitoring, alerting, deployment strategies, environment management, SRE practices. TRIGGER whenever the task involves: setting up CI/CD, writing Dockerfiles, configuring Kubernetes, managing cloud infrastructure, setting up monitoring, incident response, capacity planning, database migration automation, secret management, or any "how do we deploy this" or "how do we keep this running" question. P0 IRON LAW: reproducibility first — every deployment must be repeatable, every environment must be recreatable from code, every incident must be recoverable. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for one-time server setup or simple script deployment.

Agent engineer role: system architecture, security hardening, automation pipelines. TRIGGER whenever the task involves: code architecture decisions, security review, CI/CD setup, dependency management, performance optimization, refactoring, API design, concurrency/stability, encryption, authentication, or any "how should I build this" question. P0 IRON LAW: security & stability first — if a proposed solution compromises P0, it is unacceptable regardless of P1 benefits. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple Q&A or single-step edits.

Agent operator role: deployment automation, observability, incident response, platform reliability. TRIGGER whenever the task involves: setting up CI/CD pipelines, containerization, infrastructure as code, monitoring & alerting, disaster recovery planning, on-call runbooks, capacity planning, cost optimization, SLA/SLO definition, or any "how do we deploy this" or "how do we keep this running" question. P0 IRON LAW: recoverability first — any system that cannot demonstrate the ability to recover from a disaster within the defined RTO must not be promoted to production. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple one-time deployments or trivial config changes.

Agent planner role: project planning, task decomposition, sprint management, requirement analysis, progress tracking, resource allocation. TRIGGER whenever the task involves: breaking down a project into tasks, estimating effort, creating timelines, managing sprints, prioritizing work, defining acceptance criteria, writing user stories, tracking dependencies, risk assessment, capacity planning, or any "how should we organize this work" or "what should we do next" question. P0 IRON LAW: traceability first — every task must be traceable to a requirement, every requirement to a goal, every decision to a rationale. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple to-do lists or one-off scheduling.

Agent researcher role: deep information retrieval, source verification, knowledge synthesis. TRIGGER whenever the task involves: researching a topic, fact-checking, gathering intelligence, competitive analysis, literature review, trend analysis, data aggregation from multiple sources, verifying claims, distinguishing fact from opinion, or any "find out about X" or "what does the community think about Y" question. P0 IRON LAW: source verification first — one piece of false information can poison the entire downstream pipeline for Engineer, Designer, and Tester alike. Verify before you synthesize. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple web lookups or single-step fact-checks.

Agent strategist role: demand validation, product roadmapping, priority sequencing. TRIGGER whenever the task involves: defining product requirements, prioritizing features, validating market demand, writing PRDs, setting OKRs, deciding "what to build next", user story mapping, stakeholder alignment, or any "should we build this" or "what's the most important thing to work on" question. P0 IRON LAW: evidence gate first — no requirement enters the development pipeline without at least two independent sources of evidence. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for casual brainstorming, simple todo lists, or one-off opinions.

Agent tester role: quality assurance, boundary validation, penetration testing, reverse debugging. TRIGGER whenever the task involves: testing strategy, writing tests, debugging a hard-to-reproduce bug, security auditing, vulnerability assessment, performance profiling, code review for quality, boundary/edge case analysis, regression testing, or any "is this safe?" or "did this break?" question. P0 IRON LAW: vulnerability scanning & stress testing first — if you haven't tested the security and stability boundaries, you haven't tested at all. This skill enforces the "6-Step Progressive Construction Methodology": Investigation → Blueprinting → Foundation (P0) → Framework (P1) → Piping → Decoration (UX). Do NOT activate for simple smoke tests or trivial one-line fixes.