Shape agent capabilities into your workflow.

Add Vibe Security to a compatible coding agent to review fast-built applications for hardcoded secrets, weak database policies, broken authorization, unsafe payment flows, exposed AI keys, and related deployment risks. It loads references according to the technologies it detects, such as Supabase, Stripe, or React Native, and can activate around sensitive code. Its guidance complements testing and human review rather than replacing them.

CLI Agent Orchestrator places real coding CLIs in isolated tmux sessions and lets a supervisor delegate work through MCP. Workers can run in parallel, in sequence, or as a swarm, while the CLI, bundled web UI, and management server expose the same coordination model across providers.

Parallel Code creates a branch and Git worktree for each task, then launches Claude Code, Codex, Gemini, or another supported coding CLI inside the isolated directory. Its desktop interface keeps sessions, progress, and diffs together, supports inline review comments and head-to-head comparisons, and lets you merge selected results back into the main branch. A Git repository and at least one compatible CLI are required.

Claude Code Skill Factory provides guided agents, templates, and examples for creating Claude Skills, Claude Code agents, slash commands, hooks, and reusable prompts. Its workflow can choose between prompt-only and coded outputs, then add validation, documentation, examples, and ZIP packaging. A Codex CLI bridge also helps translate and synchronize agent guidance between tools, making the repository useful for teams producing many custom agent components.

This dashboard connects to Claude Code's native hooks, records session and tool events in local SQLite, and streams updates to a browser or macOS and Windows desktop app over WebSockets. It surfaces active agents, tool usage, health metrics, notifications, activity feeds, and Kanban-style status, making it easier to observe complex subagent orchestration as it happens.

This collection gives Claude Code reusable workflows for MCP connections, presentations, programmatic video, SOPs, skill creation, and consistent brand voice. Detailed guidance loads only when a task calls for it, keeping the working context focused. Some skills work immediately, while MCP, presentation, and Remotion workflows require their own configuration or project setup.

FlyAI connects Claude Code, OpenClaw, and other compatible agents to Fliggy travel inventory. Users describe dates, destinations, and preferences in Chinese or English, then receive structured JSON for flights, hotels, attractions, concerts, and related options. Agents can filter, compare, or render the results, and items may include direct booking links. Basic use works without configuration, while an optional API key enables enhanced access. The workflow requires both the skill and its Node.js CLI, so it is not a standalone booking app.

Atom is a self-hosted platform for running business workflows from text or voice requests. Specialized agents plan and execute work across connected services, while visual workflows, memory, and governance levels help operators control how tasks progress. Workflow data, agent state, and memory remain on infrastructure controlled by the operator. Prompts still go to a configured cloud model when bring-your-own API keys are used, unless local models such as Ollama or Llama.cpp are selected. Deployment requires operating the backend, frontend, database, and model configuration.

LLM SAST Skills turns a compatible coding assistant into a coordinated static-analysis workflow. It first maps the stack, architecture, entry points, data flows, and trust boundaries, then runs thirteen vulnerability-focused skills in parallel to discover and verify issues such as SQL injection, XSS, RCE, SSRF, authorization flaws, path traversal, unsafe uploads, and business-logic bugs. A final skill consolidates evidence, remediation guidance, and dynamic test instructions by severity. Existing CLAUDE.md or AGENTS.md files may conflict with its orchestrator.

Agentic SEO Skill routes website audits, technical SEO, content review, structured data, sitemaps, images, and GEO work to specialized skills and agents. Users can request a focused check or a full-site audit in natural language, while optional scripts collect page, crawl, performance, and metadata evidence for reports and prioritized action plans. It installs in native formats for Claude Code, Codex, Cursor, Windsurf, and several other agent environments; some collectors require Python packages or external services.

SciWrite applies a systematic editorial process derived from Kristin Sainani’s Writing in the Sciences course to scientific and engineering manuscripts. It reviews issues such as clutter, passive or weak verbs, and sentence structure in sequential passes, then suggests clearer alternatives. The skill can be installed in Claude or adapted to other tools that accept reusable instructions.

hue is a design skill for Claude Code and Codex that analyzes a brand from a URL, name, or screenshot, then turns the findings into a reusable coding skill. The generated system defines color, typography, spacing, components, light and dark modes, and page composition so separate AI sessions can build more consistent interfaces. A validation script checks the result for broken token references, placeholders, contrast issues, and structural errors before it is used.

Pexo adds a multi-shot video workflow to Claude Code, Cursor, OpenClaw, and other compatible agents. It accepts text, images, webpages, scripts, or audio, then selects models, plans shots, creates assets, and assembles voiceover, music, subtitles, transitions, and lip sync into a finished video. The workflow requires a Pexo API key.

This collection provides domain-focused Claude skills for security, privacy, legal, and governance work across frameworks such as ISO 27001, SOC 2, GDPR, HIPAA, FedRAMP, PCI DSS, and the EU AI Act. The skills support gap assessments, first-draft policies, control narratives, evidence preparation, document review, and technical recommendations tied to specific regulatory requirements.

VT Code is a Rust coding agent for long-running automation, available through an interactive TUI or headless commands and compatible with more than 21 cloud and local model providers. Restricted shell sandboxing, tool guardrails, and audit logs define its execution boundaries, while worktree isolation and Skills/MCP support extend workflows. Local inference through Ollama, LM Studio, and llama.cpp is still marked experimental.

Groq Code CLI is an open coding CLI designed as a compact, modifiable starting point rather than a feature-heavy framework. It keeps familiar terminal interactions, tools, and command structure while making it easier to add custom tools, slash commands, or UI changes. It runs on Groq models and requires a Groq API key.

Marketing Skills offers more than 160 Markdown-based playbooks for technical SEO, keyword research, page creation, paid ads, channels, launches, and analytics. You can install only the skills needed for a task in Cursor, Claude Code, OpenClaw, or other compatible agents. Tailored results depend on supplying project context such as the product, audience, brand, and target keywords.

kicad-happy gives Claude Code, Codex, and other agents workflows for parsing KiCad projects, checking nets, voltages, layouts, signal integrity, and EMC risks, and preparing component, BOM, and fabrication work. SPICE analysis requires a compatible simulator such as ngspice, LTspice, or Xyce. Distributor lookups can use optional API keys and otherwise fall back to web search.