High-priority development skills for planning, GitHub workflow, code review, testing, docs, and delivery operations.
Use this collection when software delivery is slowed down by gaps in planning, review, testing, documentation, or operational follow-through.
Supports developer workflows across planning, coding, review, testing, documentation, and DevOps delivery work.
Use this as the primary planning skill because it is the clearest starting point for structuring a feature request.
View install pathEngineering teams, technical leads, and solo builders who want a tighter software delivery loop.
Prove the main workflow first, then add alternatives only when they change the outcome.
A team wants a tighter path from feature brief to tracked implementation work.
Grouped by capability module. Numbers indicate the recommended install order. Start with primary skills before adding alternatives.
Turn product requests into clearer briefs, decomposed tasks, and implementation-ready plans.
Use this as the primary planning skill because it is the clearest starting point for structuring a feature request.
Multi-source research is more thorough than single-document mining.
Use this when vague prompts are causing rework and the team needs a repeatable constitution -> specification -> plan -> tasks -> implementation loop.
Find likely defects faster and narrow failures without relying only on a generic coding assistant.
Provides a comprehensive security checklist vs. generic code quality review.
Forces a hypothesis-first approach to debugging rather than guessing fixes.
Use this when the team wants several composable engineering motions from one maintained skill pack rather than a heavyweight end-to-end methodology that takes over planning and implementation.
Use this when the team's risk is not missing another prompt, but installing or writing skills without a review workflow for permissions, scripts, prompt injection, secret exposure, and supply-chain behavior.
Strengthen automated testing and real-browser confidence before release.
Provides proven E2E patterns rather than generic test generation.
Adds real-browser operation when the bottleneck is reproducing or inspecting a live UI flow, not designing the whole E2E test suite.
Use this when a local web app needs a repeatable reconnaissance-then-action testing workflow, especially before turning a manual UI check into a durable E2E test.
Use this when the team already trusts Playwright and wants browser access exposed as a standard MCP server rather than a one-off testing skill or bespoke automation script.
Use this when browser-agent work needs a concrete domain policy and CDP-gated tool boundary before the team trusts autonomous browsing or scraping flows.
Produce clearer API and codebase documentation alongside implementation work.
Keep as alternative when the need is documentation quality rather than API design decisions.
Use this when the API work must become an agent-usable MCP server with schemas, tool descriptions, error handling, and evaluations.
Use this when code generation depends on fast-changing APIs and the agent needs fresh, version-aware documentation instead of relying on training data or manual tab switching.
Use this when the agent needs structured database access through a maintained MCP server instead of ad hoc credentials, pasted SQL snippets, or one-off shell scripts.
Use this when documentation work should become a tested SKILL.md workflow with clear instructions, examples, and evaluation prompts.
Generates visual diagrams (flowcharts, sequence, C4) that communicate system design better than prose.
Support rollout, infrastructure, and monitoring when shipping is blocked beyond code itself.
Use this when the delivery bottleneck is choosing the right Claude Code automation surface for a specific codebase rather than adding another generic coding assistant.
Use this when the delivery question is not just writing agent code, but mapping it onto Cloudflare's runtime primitives: Durable Objects state, callable RPC, queues, retries, workflows, MCP transport, and observability.
Use this when the operational question is how to choose, configure, test, and deploy Microsoft cloud or agent-platform code with skills, MCP server configs, agent personas, prompts, and AGENTS.md guidance in one official repository.
Use this when the team wants Codex-specific skills with an official install path, a visible split between built-in system skills and opt-in curated skills, and source folders that can be reviewed before adoption.
Use this when the agent's operational work depends on Elastic-specific APIs, ES|QL, Kibana objects, SLOs, alerts, LLM observability, or security cases instead of generic log-search or monitoring prompts.
Use this when the delivery work depends on Browserbase infrastructure and the agent should stay inside the official bb CLI instead of switching to ad hoc browser scripts.
Keep issues, PRs, and execution tasks aligned inside a GitHub-driven delivery loop.
Covers the full GitHub workflow loop rather than just issue creation.
A collection should explain who it serves, what it solves, and what to install first. This section makes the decision criteria explicit before installation.
A team wants a tighter path from feature brief to tracked implementation work.
A maintainer needs stronger review and testing before merge.
A product team needs docs and delivery operations to keep up with shipping speed.