Claude Devstudio

Intelligent test execution based on context with automatic failure analysis and fixes

Record, list, search, and apply cross-session learnings — capture insights from completed work as JSONL entries so patterns persist across conversations and accumulate into institutional knowledge. Inspired by gstack's learn skill.

Run a weekly retrospective — what shipped, what broke, what slowed you down, and what to change next week. Produces a dated retro entry and concrete action items.

Monitor a deployment for errors and regressions after shipping — polls logs, error rates, and key endpoints in a configurable loop until the deployment is confirmed healthy or a problem is detected.

Analyze changes and create meaningful conventional commits with pre-commit quality checks

Investigate LLM output quality issues — trace hallucination root causes across context, retrieval, prompts, tool use, and workflow design

Multi-agent code analysis covering security, performance, quality, and architecture

Synthesize the current conversation context and codebase into a full PRD — problem statement, user stories, implementation decisions, testing decisions, and scope boundaries. Use when user wants to turn a discussion into a formal PRD or publish requirements as an issue.

Move issues through a state machine with category and state labels — classify as bug or enhancement, assign triage state, reproduce bugs, and track out-of-scope requests. Use when triaging GitHub issues, processing a backlog, or managing issue lifecycle.

Intercept destructive commands before execution — rm -rf, DROP TABLE, force-push, git reset --hard, and similar irreversible operations. Prompts for confirmation and suggests safer alternatives. Inspired by gstack's careful skill.

Activate ultra-compressed communication mode that drops ~75% token usage by eliminating articles, filler, pleasantries, and hedging while preserving technical accuracy. Use when user says "caveman mode", "be brief", "less tokens", or invokes /caveman.

Generate API mocks and stub servers from OpenAPI specs or code analysis

Generate a daily development briefing — recent git activity, open TODOs, pending PRs, failing tests, and session context — so you can start work immediately without manually checking multiple sources.

Configure CI/CD pipelines for GitHub Actions, GitLab CI, CircleCI with best practices

React to CI failures on main or active branches — diagnose the failure type, propose a minimal fix for regressions, re-run flakes, and escalate infrastructure failures. Designed to run with /loop 15m.

Build a queryable knowledge graph from the current codebase using the Graphify CLI — code, SQL schemas, docs, and more in one interactive graph

Relentlessly interview the user about a plan or design, walking down every branch of the decision tree one question at a time until reaching shared understanding. Use when user wants to stress-test a plan, get challenged on their design, or says "grill me".

Convert a bug fix, repeated workaround, or solved problem into a permanent, testable Claude Code skill. Runs a 10-item conformance checklist before writing the skill file.

Design planning interview before writing UI code — produces a design brief covering register, typography strategy, color constraints, and motion budget to prevent costly rework

Apply Domain-Driven Design — identify bounded contexts, model aggregates and value objects, map context relationships, and generate implementation scaffolding

Proactively discover outdated and vulnerable dependencies, apply patch and safe minor updates in isolation, verify with tests, and escalate major bumps and high-severity CVEs to a human. Designed to run with /loop 6h or /loop 1d.

Sync documentation files with code changes — detects what changed, finds affected docs, and applies surgical updates so docs never drift from code. Works in any repo.

Import and adapt code from any source to your project with validation and testing

Implement lazy loading patterns for images, components, and routes

Generate realistic seed/fixture data based on schema analysis

Production-readiness pass for UI — audits and fixes error states, empty states, text overflow, small touch targets, and i18n gaps before shipping

Fix spacing rhythm, visual hierarchy, and layout issues in UI files — enforces consistent spacing units, adequate padding on interactive elements, and proper visual weight distribution

Final pre-ship UI quality pass — reads the persisted critique snapshot as a prioritized backlog and implements fixes from P0 down

Analyze codebase for resilience gaps — missing retry logic, absent timeouts, circuit-breaker opportunities, single points of failure, and degradation-mode blind spots

Report token spend per session and cumulative totals from session logs

Run Lighthouse audits and implement performance, accessibility, and SEO fixes

Sweep for follow-up work after recent merges — TODOs added in merged PRs, stale feature flags, dead code, documentation gaps, and tech-debt tickets left by merge comments.

Run 8 post-session-start health checks to verify the project is in a working state — git clean, no broken imports, tests pass, key files present, and agent memory intact.

Review code and configuration for SRE concerns — SLO/SLA coverage, observability gaps, toil, alerting blind spots, and reliability anti-patterns

Async testing patterns with race condition and timing issue detection

Add purposeful animations and transitions to UI — enforces motion budget, bans bounce/elastic easing, and requires prefers-reduced-motion fallbacks on every animated element

Fix typography hierarchy, font choices, type scale, and line-length in UI files — applies register-aware rules (brand vs product) with concrete CSS fixes

Project architecture analysis to understand structure, patterns, and dependencies

Create detailed implementation plans with task breakdown

Generate framework-specific boilerplate code

Repair broken imports automatically across the project

Generate JSDoc/docstrings from code analysis

Pre-flight readiness checklist for a new loop before enabling unattended execution — verifies maker/checker split, state file, budget caps, denylist, kill switch, attempt cap, and human escalation paths

Independent verification agent for loop-produced changes — runs tests, confirms diff scope, and issues APPROVE / REJECT / ESCALATE_HUMAN. Use after any implementer agent. Never in the same role as the implementer.

Produce the smallest possible code change that fixes one specific, well-scoped issue — CI failure, reviewer comment, or bug report. Never touches unrelated code and respects a safety denylist.

OWASP Top 10 vulnerability scanning and remediation

Introduce strategic color to monochromatic or under-colored UIs — OKLCH-first palette generation with contrast validation, register-aware strategy, and specific anti-pattern warnings

Dual-agent UX design review — Assessment A runs design-director heuristics while Assessment B runs visual pattern analysis; results synthesized into a prioritized critique snapshot

Generate API usage examples and tutorials from code analysis

Remove debug artifacts and temporary files safely with git checkpoint protection

Save current work-in-progress context as a git WIP commit — captures all staged and unstaged changes with a timestamped stash message so you can switch contexts safely and restore later. Inspired by gstack's context-save skill.

Contribution readiness assessment analyzing project guidelines and requirements

Record architectural and design decisions as structured ADR files in docs/decisions/, with auto-indexing and link suggestions to related decisions

Audit the developer experience of your project — setup friction, onboarding clarity, local dev loop speed, tooling consistency, and documentation gaps. Produces a DX scorecard and prioritized improvement list. Inspired by gstack's devex-review skill.

Use when the user requests diagrams, flowcharts, architecture diagrams, ER diagrams, UML / sequence / class diagrams, network topology, cloud architecture from Terraform or Kubernetes manifests, ML/DL model figures (Transformer/CNN/LSTM), mind maps, or any visualization. Also use proactively when explaining systems with 3+ components, complex data flows, or relationships that benefit from visual representation. Best suited when the diagram needs custom styling, rich shape vocabulary, swimlanes, or exportable images (PNG/SVG/PDF/JPG). Generates .drawio XML and exports locally via the native draw.io desktop CLI.

Generate end-to-end tests with Playwright browser automation

Execute implementation plans in controlled batches with validation and checkpoints

Manage git worktrees for parallel feature development

Extract requirements one targeted question at a time before any plan or code starts — hypothesizes user intent, scores confidence, and stops at ~95% confidence.

Discover, deduplicate, prioritize, and label incoming GitHub issues and discussions — produce a structured priority queue with one-line summaries and suggested labels. Designed to run with /loop 2h or on demand.