multi-agent-patterns: Multi-Agent Collaboration Architectures and Context Isolation
multi-agent-patterns, from Agent-Skills-for-Context-Engineering, targets multi-agent system designs that require context isolation, supervisor or swarm coordination, explicit handoffs, and parallel execution. The skill emphasizes that sub-agents should primarily be used for context isolation rather than role-play, and summarizes three patterns—Supervisor/Orchestrator, Peer-to-Peer/Swarm, and Hierarchical—along with the tradeoffs involved in selecting among them. It covers the telephone-game problem and direct forwarding via forward_message, includes cautions about token consumption relative to single-agent designs, and clarifies boundaries with adjacent skills such as project-development and hosted-agents. It is suitable for developers making architecture decisions and running baseline evaluations in production-grade, multi-component agent systems.
Read MoreSelf-Improvement Loops: When the Harness Itself Becomes the Optimization Target
self-improvement-loops is a Skill newly introduced in Agent Skills for Context Engineering 2.4.0, targeting scenarios where the harness, workflow, or optimizer itself becomes the optimization target. The article explains its optimization ladder, out-of-loop evaluation invariants, empirical acceptance gating based on held-in/held-out data, filesystem-backed experience archives, and failure-driven three-stage self-editing, and provides directory-based installation instructions and official examples for Cursor, Claude Code, and Codex, helping it coordinate with adjacent Skills such as harness-engineering and evaluation.
Read Moredsh-model-health-probe: DSH Model Health Check Probe
The DeepSeek Harness (DSH) adds a new “Model Health Check” tab, aimed at addressing the difficulty of pinpointing configuration errors. The plugin supports using a four-line selector to dynamically load providers, protocols, models, and routes from `settings.yaml` for testing. It provides raw HTTP request capabilities, supports the OpenAI and Anthropic protocols, and offers streaming/non-streaming switching, a three-layer diagnostic chain, and Chinese error messages. Users can configure either strict or connectivity retry modes and test request latency, TTFT, and HTTP status. The plugin is installed via a local `.tgz` package and relies on host configuration and credential services, but does not record historical trends or monitoring alerts.
Read MoreHarness Engineering: Designing Runtime Frameworks for Autonomous Agents
harness-engineering is a skill within Agent Skills for Context Engineering, used to design the runtime harness for autonomous agents: it distinguishes locked, editable, append-only, and human-controlled surfaces; establishes tight feedback loops and persistent THREAD logs; pairs novelty gating, pruning, rollback, and PR preparation; and clarifies human approval boundaries for merge and deployment. Based on the official SKILL.md, the article outlines the Autoresearch and Research-To-Skill loops, key anti-metric-gaming considerations, and the directory layout, explaining when to enable it and how it divides responsibilities with adjacent skills such as evaluation.
Read Morebabysitting-pr: Have the Agent Continuously Keep PRs Mergeable
"babysitting-pr" is from awesome-cursor-skills, a SKILL.md that instructs the Cursor Agent to fetch status via gh when a PR is opened, address CI failures and review comments, resolve merge conflicts with main, and recheck using gh pr checks --watch after pushing. The document specifies a maximum of three fix-push-check cycles, prohibits force-pushing, and requires reporting items that need design decisions to the user. It is suitable for teams that want to keep PRs continuously merge-ready; to install it, place the skill in .cursor/skills/.
Read Morewriting-tests: A Cursor Skill for Generating Unit and Integration Tests for Existing Code
writing-tests, from awesome-cursor-skills, is a SKILL.md workflow for Cursor Agent that identifies testing stacks such as Jest/Vitest, analyzes exported APIs and uncovered branches, generates unit and integration tests according to project conventions, and uses vi.mock and Testing Library to handle dependencies and React components. It is suitable for adding missing tests, improving test coverage, or writing tests for a single module; place the Skill in .cursor/skills/ for automatic discovery by the Agent.
Read Moreadding-docker: Generate Production-Grade Docker Configurations for Your Application
adding-docker is an Agent Skill in awesome-cursor-skills. When a user asks to containerize an application, it follows the SKILL.md workflow to detect the runtime (e.g., package.json, go.mod), generates a multi-stage Dockerfile and .dockerignore, and adds docker-compose.yml when dependencies such as a database are present. The documentation emphasizes using Alpine images, never baking .env into the image, ignoring unrelated packages in a monorepo, and including a production HEALTHCHECK. Installation is done by placing SKILL.md in .cursor/skills/ so the Agent can discover it automatically.
Read MoreSuggesting Cursor Rules: When Correcting the Same Convention Repeatedly, Write It as a Cursor Rule
suggesting-cursor-rules is a Cursor-Native Skill in awesome-cursor-skills. When a user corrects the same code convention two or more times, expresses a fixed preference, or the Agent repeatedly makes the same type of error, the Agent proposes creating a .mdc rule file under .cursor/rules/. After user consent, it writes the description, globs, and convention entries. The Skill is user-invocable and requires copying SKILL.md to .cursor/skills/ so that the Agent can discover it automatically. This article introduces the trigger conditions, proposal wording, rule template, and official examples, explaining the meta-workflow in which Rules and Skills work together.
Read MoreA Plugin for Integrating DeepSeek Harness with the Command Code Subscription
The DeepSeek Harness (DSH) plugin replicates the Command Code private gateway protocol, converting subscriptions into a DSH-native provider. Core features include a real-time model catalog, context capacity annotations, visual API key configuration, and a bottom usage-ring monitor. The plugin supports streaming output, tool calls, and image input. Installation is performed via the command line with simple configuration. When using it, note that images may be resubmitted in long conversations, automatic retries are unavailable, and single-account restrictions apply.
Read Morevoltagent-docs-bundle: Version-Matched VoltAgent Queries Using Documentation in node_modules
voltagent-docs-bundle is an official VoltAgent Agent Skill that guides coding assistants to consult version-matched Markdown documentation in node_modules/@voltagent/core/docs for API signatures, guides, and examples. Rather than relying on the official website or training data, the docs are bundled with @voltagent/core, reducing version drift. The article explains the documentation path, the ls/rg/cat lookup workflow, the installation method via npx skills add VoltAgent/skills, as well as applicable scenarios and prerequisites.
Read Morememory-systems: A Design Guide for Agent Persistent Memory and Retrieval Architectures
memory-systems is the Memory System Design guide in the Agent Skills for Context Engineering collection, covering cross-session semantic memory, entity consistency, temporal validity, graph/vector and hybrid retrieval, and memory consolidation. The article explains its positioning, activation boundaries, memory hierarchy and framework selection table, and principles for consolidation and JIT loading. It also provides official examples using Mem0, temporal graphs, and Cognee, along with loading methods via Claude Code and npx skills, emphasizing starting with a shallow design, validating it with LoCoMo/LongMemEval, and then deepening the architecture.
Read Morecontext-compression: Context Compression and Token Optimization for Long-Session Agents
context-compression, from Agent-Skills-for-Context-Engineering, focuses on context compression and handoff summaries for ultra-long conversations. The article sets the optimization target as tokens-per-task, introduces three paths—anchored iterative summarization, opaque compression, and full-summary regeneration—emphasizes artifact trails and mandatory sectioned summaries, and provides trigger strategies, probe-based evaluation methods, and common enablement pathways such as Claude Code and npx skills.
Read Morefirecrawl-agent: Enabling AI to Autonomously Crawl Websites and Extract Structured Data Based on a Schema
firecrawl-agent is an Agent Skill included with the Firecrawl CLI, teaching coding assistants to run firecrawl agent in the terminal: autonomously crawl sites, collect data across pages, and constrain output with a JSON Schema. The article explains how it complements Alexandria’s ready-made workflows, covers installation and authentication (npm / init / setup core), outlines typical commands (--wait, --schema, --urls, job polling), and discusses the trade-offs between credits and scrape, making it easier to integrate web data into AI application pipelines.
Read MoreAudit Context Building: Build the Audit Context Before Vulnerability Hunting
Trail of Bits' audit-context-building Skill begins the audit and threat modeling process by documenting assumptions, guarantees, and dependencies on a per-function basis. It generates a DOSSIER and function-level records, marking “nothing found” results and open questions, and does not name vulnerabilities. It is compatible with Claude Code, Codex, and similar tools, and can be installed via the trailofbits/skills marketplace.
Read Moredsh-tavily-resilient-search: A Resilient Tavily Web Search Plugin for DSH
`dsh-tavily-resilient-search` is a standalone plugin in the DeepSeek Harness (DSH) ecosystem designed to address Tavily API rate limiting, quota exhaustion, and key invalidation issues. Core features include atomic key rotation, a Keyless fallback mode, and a web admin panel. Installation requires Node.js 24+ and supports configuring the key pool via environment variables. The plugin provides robust error handling, automatically switching keys or falling back to maintain continuous Agent loops, and returns structured search tool data.
Read MoreProduct Strategy: Building a Product Strategy with the Nine-Section Product Strategy Canvas
product-strategy is an Agent Skill under the pm-product-strategy plugin in phuryn/pm-skills. It guides the assistant to create a strategic document using the nine sections of the Product Strategy Canvas—vision, segments, costs, value proposition, trade-offs, metrics, growth, capabilities, and defensibility—and performs consistency checks, identifies key assumptions, and recommends low-effort validation experiments. It supports Claude Cowork/Code, Codex CLI, and environments like Cursor by copying SKILL.md. Its methodology shares the same origin as Product Compass and is suitable for product teams that need structured product direction.
Read Moreprd-development: An Agent Skill for Turning Discovery Notes Into a Deliverable PRD
prd-development from deanpeters/Product-Manager-Skills is a workflow-style Skill for major requirements. It uses a 10-section PRD template and eight stages to weave together the problem, users, strategic context, solution, success metrics, user stories, and out-of-scope items into a living document, and marks gaps in template.md with Assumption/Open Question notes. It can be invoked with discovery materials or started from problem definition, and can be paired with workshop-facilitation for step-by-step guidance, making it suitable for structuring requirements before engineering handoff.
Read MoreCopywriting: Using Agent Skill to Write Marketing Copy for Landing Pages and Pricing Pages
copywriting is an Agent Skill in Corey Haines' marketingskills repository (v2.3.1) for writing and optimizing marketing copy for homepages, landing pages, pricing pages, and similar pages. It first reads the product-marketing context, then diagnoses one of three cases: drafting from scratch, preserving tone, or making targeted edits. It drafts headlines and CTAs using decomposition locking and a three-question gate, and rejects unsupported claims. The full repository can be installed with npx skills, and it works with tools that support Agent Skills, such as Cursor, Claude Code, and Codex.
Read MoreReact Email Skill: Write HTML Emails with React Components and Integrate Resend Workflows
react-email is an Agent Skill (version 2.1.0) provided by Resend in resend-skills. It guides AI systems to write HTML email templates using React Email, preview them locally, handle static assets, and support transactional email scenarios such as welcome messages, password reset, notifications, and order confirmation. The article explains the relationship between this Skill and the react-email open-source project, how to install it using `npx skills add resend/resend-skills`, integrate it into a development environment with `npm/create-email`, and apply constraints related to Tailwind, pixelBasedPreset, email client layout, and images, enabling reproducible use in clients such as Cursor.
Read Morewp-playground: Guiding AI Assistants Through WordPress Playground, Scenario by Scenario
wp-playground is the Playground routing wrapper in the official WordPress agent-skills. Instead of duplicating the Blueprint schema, it routes by intent to the blueprint skill or references such as cli, website, and debugging. It covers local execution with @wp-playground/cli, sharing on playground.wordpress.net, mounting, version switching, and Xdebug and WebMCP browser workflows. It requires Node.js 20.18+ and targets WordPress 7.0+. It can be installed into a project or the global Cursor directory via npx skills add.
Read MoreDeepSeek Harness Remote Access Plugin (dsh-guard-messenger)
`dsh-guard-messenger` is a Cordis plugin for DeepSeek Harness that provides a secure remote access solution through a Telegram Bot and Cloudflare Tunnel. The service runs on the local address `127.0.0.1`, exposes only a key-authenticated proxy entry point, and supports passwordless access, tunnel toggling, and key rotation. This plugin is suitable for accessing the Web UI to perform coding from mobile devices in restricted network environments. Note that content is transmitted over Cloudflare; for extremely high security requirements, using SSH or Tailscale is recommended.
Read Moresanity-best-practices: An Agent Skill Guide for Full-Stack Development with Sanity
sanity-best-practices is maintained by Sanity in the agent-toolkit, covering schema design, GROQ, TypeGen, Visual Editing, localization, migrations, Functions, Blueprints, and framework integrations such as Next.js. The Skill provides global rules (such as document _id and video assets) and topic-specific references, supporting loading only 1–2 guides per task. It can be installed via `npx skills add sanity-io/agent-toolkit`, and can also be used with MCP (`npx sanity@latest mcp configure`) or Cursor/Claude Code plugins.
Read Morefastify-best-practices: Aligning Fastify Backend Practices with Agent Skills
fastify-best-practices is located at skills/fastify in mcollina/skills and was released by Matteo Collina, a core maintainer of the Fastify ecosystem, to unify Fastify and REST API practices during AI-assisted development. The skill covers topics such as plugin encapsulation, schema validation, error handling, performance, and security through rules files for plugins, routes, schemas, hooks, authentication, testing (inject), logging (Pino), TypeScript, and deployment, and provides scenario-based reading orders along with minimal runnable examples. This article explains its positioning, capability boundaries, and the general ways to load it through platforms such as npx skills and Claude Code.
Read MoreExpo-Upgrade: Using the Official Skill to Turn Expo SDK Updates Into a Reproducible Process
expo-upgrade is an MIT-licensed open-source Skill maintained by the Expo team in expo/skills, designed for Expo SDK upgrades, dependency conflicts, deprecated package migrations, and cache cleanup. The article provides step-by-step commands (expo install, expo-doctor, cache clearing, and prebuild branching), SDK 53–56-specific references, notes on Hermes and the New Architecture, and a housekeeping checklist. It can be installed in Cursor and similar tools using npx skills add expo/skills; Claude Code and Codex can use the official Expo plugin. It is well suited for AI-agent-driven reproducible upgrades and reducing missed breaking changes in peer dependencies and packages such as expo-av.
Read Moreeas-workflows: Enabling Agents to Write EAS Workflow YAML According to the Specification
eas-workflows is an Agent Skill maintained by the Expo team, intended for configuring EAS CI/CD under `.eas/workflows/`. It requires the Agent to first use a fetch script to retrieve the workflows JSON Schema, syntax, and preset Job documentation before modifying YAML, and then run `npx eas-cli workflow:validate` to perform server-side validation against `eas.json`. The Skill declares Read/Write and EAS CLI-related Bash permissions; EAS is a paid service. Cursor and similar tools can install it via `npx skills add expo/skills`.
Read More