Agent Automation & Testing Skills

Use to ASSESS, HARDEN, SCAFFOLD, or DIAGNOSE a repository's agent-readiness for coding-agent harnesses such as Claude Code, Cursor, Codex, Copilot, Windsurf, and Aider.

Use when you operate coding-agent infrastructure with a feedback signal — eval suites, run-level telemetry, A/B baselines, or a skill catalog under test — and want to…

Use to DESIGN or MEASURE documentation — plan or reshape a docs/help/agent-readable surface before implementation (information architecture, mode taxonomy, README/quickstart/reference structure, examples strategy, API/tool-contract docs, error-copy shape, versioning, accessibility baseline, agent context files), or define the telemetry, CI gates, sample tests, retrieval evals, feedback loops, and success metrics that keep docs true. Triggers on "design our docs IA", "how should we structure the quickstart/reference", "what should our tool descriptions look like", "how do we measure docs quality", "set up docs CI gates / freshness checks". Emits a design doc with a concrete proposed structure and acceptance criteria, or a measurement plan with signals, thresholds, owners, and actions. Do NOT use to AUDIT or debug existing docs (use docs-audit), for end-user visual UI design (use ux-audit), or for repo-level agent hardening / AGENTS.md design (use agent-experience).

Use to AUDIT an existing developer-experience surface — audit it for friction and score it, debug a reported developer-facing issue, or run a pre-ship edge-case risk pass. Covers APIs, SDKs, CLIs, developer docs, examples, errors, setup, local inner-loop, migrations, package contracts, contributor flows, auth, IDE integration, plugins, perf perception, telemetry, README/first-impressions, logging, config, and changelogs. Triggers on "DX review", "audit our API/SDK/CLI", "is this error message OK", "why do users hit this", "pre-1.0 risk pass". Do NOT use to DESIGN a new surface from scratch (use dx-design), for end-user product UX/accessibility (use ux-audit), or for AI-agent-facing surfaces, AGENTS.md, or AI/Agent SDK design (use agent-experience).

Use to DESIGN or shape a new developer-experience surface from scratch — an API, SDK, CLI, developer-doc IA, error envelope, setup/first-run flow, auth model, migration/deprecation path, plugin or extension contract, package/publish scheme, logging/config design, changelog convention, or telemetry/consent contract. Triggers on "design a new API/SDK/CLI", "what should we get right up front", "shape the public surface before we write code", "what should our error envelope look like". Emits a design doc with a concrete good-shaped pattern and acceptance criteria. Do NOT use to AUDIT or debug an existing surface (use dx-audit), for end-user visual UI (use ui-design), or for AI-agent-facing surfaces / AI/Agent SDK design (use agent-experience).

Use when evaluating, designing, or debugging developer experience for APIs, SDKs, CLIs, docs, examples, setup, errors, local dev, build/test workflows, migrations, package contracts, contributor workflows, auth, IDE integration, plugins, performance, or telemetry. Also trigger for DX reviews, developer onboarding friction, confusing integration steps, dev-facing interface design, or PR feedback about developer usability.

Use when building or operating coding-agent infrastructure where the feedback signal exists (eval suites, run-level telemetry, A/B baselines, skill catalogs under test) and you want to capture observed agent failures, promote recurring patterns into rules, and score advanced (Level 4–5) agent-readiness. Triggers on "reflection log," "agent failure log," "evidence-driven scaffolding," "W1 ≥3 floor," "promote pattern to rule," "post-incident eval case," "harness maturity Level 4 / Level 5," "Sovereign Engineering," "Spec Architecture," or "set up a feedback loop for agent failures." Pair with project-agentification for project-context-first AGENTS.md scaffolding — this skill layers the evidence-driven workflow on top.

Template contract for this repository: the minimum artifacts every skill must ship (SKILL.md, skill.json, evals/) so static checks catch a skill templated from it that bypasses gates. Hidden from npx skills add . --list by metadata.internal: true. Do not delete.

Use when researching books, movies, talks, articles, or notes as source material and turning them into source-inspired public skills.

Use when reviewing or editing proposed or existing public skills for structure, metadata, source safety, usefulness, validation, and release readiness. Do not use for building a new skill from source material; use informed-skill-curator for that.

Use to DESIGN or optimize the performance and observability of a production or runtime system — design instrumentation, SLOs/error-budget policy, tracing topology, latency budgets, or metric selection up-front for a new or expanding system; plan a profile-first optimization pass on a known-slow path with before/after measurement gates; or shape a program-level observability and reliability strategy across logs, metrics, and traces. Triggers on "design our SLOs", "what spans/metrics should we emit", "plan a tail-latency optimization", "cut our log bill without losing recovery value", "observability roadmap". Emits a design doc, optimization plan, or strategy doc with concrete good-shaped patterns and acceptance criteria. Do NOT use to AUDIT or diagnose an existing system (use perf-audit), for developer-facing or local build/inner-loop performance (use dx-design), or for end-user product UX (use ui-design to build it, ux-audit to audit it).

Use when assessing a repository's agent-readiness, hardening it against agent failure modes, scaffolding AGENTS.md / SKILL.md / MCP servers / hooks / specs from observed failures, or diagnosing why coding agents trip on this repo. Also trigger for AGENTS.md review, agent-native repo design, harness engineering, MCP exposure, agent eval setup, sandbox/approval design, context-rot debugging, repository maturity assessment, or "make this repo work better with Claude Code / Cursor / Codex / Copilot / Windsurf / Aider.

Use when researching books, movies, talks, articles, or notes as source material and turning them into draft source-inspired public skills.

Use to PRODUCE test-suite artifacts — author a new test or test plan for a feature, shape a cross-layer test strategy (what to test at which layer, as a portfolio), or plan which tests to delete (low-value, redundant, characterizing dead code). Covers unit, integration, e2e/UI, exploratory, property-based, contract, snapshot, mutation, and performance tests. Triggers on "write tests for this feature", "what should I test and at which layer", "design our test strategy", "which tests should we delete", "is our test pyramid right". Do NOT use to REVIEW or triage existing tests for smells, flakiness, or false-pass (use test-audit), or for production-system performance/SLOs (use perf-design).

Use when reviewing, designing, triaging, or rationalizing test suites — unit, integration, e2e/UI, exploratory, property-based, contract, snapshot, mutation, or performance tests. Trigger for flakiness triage, false-pass risk, brittleness on refactor, suite pruning, test-pyramid/trophy decisions, exploratory charters, and snapshot review. Routes by activity (triage / review / author / strategize / prune) and layer.

Use when designing, polishing, prototyping, or reviewing user-facing UI, frontend screens, visual systems, interaction prototypes, slide decks, motion/effects, artifact handoff, or anti-slop visual quality. Trigger for "make this UI look better", "design a dashboard", "create frontend mockup", "prototype this flow", "author a design system", "make a deck", "add motion", "prepare handoff", or "review visual craft". Do not use for pure accessibility audits without visual design work; use ux-accessibility-heuristics for that.

Use when reviewing or designing user-facing product UX, usability, accessibility, forms, navigation, error/recovery copy, onboarding flows, checkout/sign-up friction, or WCAG-oriented accessibility basics. Trigger for heuristic evaluations, cognitive walkthroughs, form friction, IA/navigation issues, keyboard/screen-reader risks, inclusive design checks, and dark-pattern reviews. Do not use for developer-facing DX surfaces; use dx-heuristics for those.

Use to AUDIT an EXISTING piece of writing — tighten wordy prose at the line and paragraph level (revise), fix mechanics, grammar, and consistency while preserving voice (copyedit), or diagnose why a draft drags, buries the point, or fails to land (diagnose). Routes by intent (revise / copyedit / diagnose) × genre (argument-memo / technical-doc / talk-pitch / narrative / general-prose). Emits a scored findings report with severity-rated findings and concrete fixes. Triggers on "tighten this", "make this clearer", "copyedit this", "why does this drag", "review my draft", "this memo buries the point". Do NOT use to write or structure something NEW (use writing-design), to audit a documentation system's IA, retrieval, or telemetry (use docs-audit), or to review product UI copy as a usability surface (use ux-audit).

Use to DESIGN or shape a NEW piece of writing before or while drafting — structure an argument or memo (PRD, RFC, recommendation, strategy), organize a task-oriented technical doc or explainer, get a first pass down without stalling, or give a talk, pitch, launch post, or narrative a persuasive arc. Routes by intent (structure / draft / persuade) × genre (argument-memo / technical-doc / talk-pitch / narrative / general-prose). Triggers on "help me outline this RFC", "draft this announcement", "make this pitch land", "give this post an arc", "structure this doc before I write it". Do NOT use to revise, copyedit, or diagnose an EXISTING draft (use writing-audit), to design a documentation system's IA, retrieval, or telemetry (use docs-design), or to gather and validate source material (use research).