Skills
100,928 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Reduce visual intensity, saturation, and noise while preserving hierarchy and character. Use when the structure is basically sound but the design feels too loud, bold, garish, aggressive, or overstimulating—not when the real problem is structural clutter, too many features, or not enough personality
Reverse engineering and static analysis of AI model files: ONNX/PyTorch/Safetensors/TFLite format parsing, network structure reconstruction, weight extraction, and file-level watermark analysis (weight patterns/metadata/tensor hashes/embedding anomalies). Trigger terms: model files, weight extraction, ONNX parsing, safetensors, pth analysis, pt files, model structure reconstruction, file-level watermarking
Reverse engineering modern consoles and retro platforms: Switch NSO/NPDM containers and encrypted partitions, PS4/PS5 ORBIS structures, SDK library fingerprints; retro ROM headers, cartridge formats, saves, and Cheat codes. Trigger terms: console reverse engineering, Switch, NSO, NPDM, PS4, PS5, ORBIS, Xbox, retro, ROM, NES, GBA, emulator, Cheat.
Firmware/embedded/hardware analysis gateway. Workflow: extraction → rootfs → emulation → hardware interfaces. Subskills: [[re-fw-extract]] [[re-fw-rootfs]] [[re-fw-emulate]] [[re-hardware-io]] [[re-automotive]] [[re-uefi]] [[re-rtos]] [[re-tee]]. Trigger terms: firmware analysis, firmware, IoT, unpacking firmware, rootfs, firmware emulation.
Memory forensics/threat intelligence gateway. Workflow: dump source → memory forensics → indicator extraction → intelligence correlation. Subskills: [[re-mem-forensics]] [[re-disk-forensics]] [[re-ti]], [[re-attribution]], [[re-hunting]], [[re-mobile-forensics]]. Trigger terms: memory forensics, memdump analysis, threat intelligence, sample context, VirusTotal, MISP, forensics, threat intelligence.
IoT protocols: MQTT/CoAP/BLE/Zigbee; BLE link layer (advertising parsing/pairing encryption) and NFC/smart cards (ISO14443/APDU/MIFARE). Trigger terms: MQTT, CoAP, BLE, Zigbee, IoT protocols, BLE link, BLE sniffing, advertising parsing, pairing encryption, GATT, NFC, smart cards, MIFARE, APDU, ISO14443
Audit and revise practical nonfiction drafts for reader experience, value pacing, front-loaded insight, value-per-page, boredom, and beta-readiness. Use when editing, restructuring, shortening, or reviewing a book, guide, manual, course-like manuscript, or long-form educational draft so readers receive useful value quickly and keep reading.
Build animated motion graphics as real video with Remotion (React-based programmatic video) — title cards, lower thirds, logo stings, badges, stat reveals, countdowns, end screens, captions, animated charts, social reels — and export them as MP4 or as transparent ProRes/PNG sequences for Premiere, Final Cut, CapCut, DaVinci or After Effects. Use this skill whenever the user wants to create, animate, restyle, polish or render any video graphic, overlay, intro, reel or animation from code, or mentions Remotion, motion graphics, video overlays, alpha/transparency export, or asks to make an existing graphic feel smoother, less flat, less robotic or more premium — even when they never say "Remotion" out loud.
Requirements-engineering expert that operates in three modes: (1) Elicitation — extract atomic, testable requirements from briefs, meeting notes, or system specs using EARS notation, with full attribute set (ID, type, priority, ASIL, verification method, source), flagging ambiguities as open questions; (2) Refinement — detect requirement defects (vague qualifiers, compound statements, implementation prescription, untestability, missing attributes) and rewrite each as a specific, measurable, EARS-formatted statement with documented assumptions; (3) Traceability — build bidirectional matrices (SYS-REQ → SW-REQ → design → test), detect orphaned requirements / design elements / tests, and flag safety gaps where ASIL-tagged requirements have no test coverage (ISO 26262-6 §9 violation). Aligned with EARS, SMART, ISO 26262-6, and ASPICE SWE.1/SWE.6 practices. Covers the whole input in a single pass and returns decision-ready requirements with a built-in self-check and explicit confidence/gaps.
Turn verified research findings into clear, calibrated scientific writing — paper sections (Results, Experiments, Discussion), lab notes, technical reports, and grant progress reports. Use this skill whenever the user has run experiments or analysis and wants help writing them up, including casual phrasings like "draft the results section", "write up these numbers", "turn my notebook into a paper section", or "summarize what we found". Also use when the user shares logs, tables, or a notebook and asks for prose. For reviewer-by-reviewer rebuttal strategy, use reviewer-response-strategist instead. The skill's main job is preventing confident-sounding overclaims and hallucinated numbers in scientific text.
Reverse engineer and intercept traffic from Flutter/Dart mobile apps using blutter, reFlutter, and Frida. Use when an APK or IPA contains libflutter.so, libapp.so, App.framework, or flutter_assets, when jadx shows only a thin Dart wrapper, when Burp or mitmproxy sees no traffic from an app that is clearly online, or when the target is described as built with Flutter or Dart.
Creates minimal, effective AGENTS.md files using progressive disclosure. Triggers on "create agents.md", "refactor agents.md", "review my agents.md", "claude.md", or questions about agent configuration files. Also triggers proactively when a project is missing AGENTS.md.
Multi-agent review-and-improve loop for a GitHub PR you have checked out: applies fix commits to the local branch, pushes them to the PR, and reports back in a PR comment. Use when the user wants to "improve a PR", "review and commit fixes", "iterate on my PR", or "review and push back" against a checked-out PR branch; the PR is auto-detected when no locator is supplied. Activates the `review-anvil` engine in per_fix mode.
Migrate from package:provider to Riverpod; ChangeNotifierProvider, ProxyProvider to ref.watch, context.watch to ref.watch, ConsumerWidget, incremental migration, family and autoDispose. Use when the user is migrating from Provider to Riverpod, or asks about Provider vs Riverpod, or how to replace ProxyProvider/ChangeNotifierProvider.
Provides comprehensive guidance for Rollup bundler including configuration, plugins, code splitting, tree shaking, and library bundling. Use when the user asks about Rollup, needs to bundle libraries, optimize output, or configure Rollup for production builds.
Connects sales tools with webhooks, Zapier/Make, native integrations, and custom API pipelines — CRM sync, event triggers, data mapping, and error handling. Use when CRM data isn't syncing from Mailshake/Apollo/Salesloft, webhook events aren't firing or arriving malformed, Zapier/Make automations are failing or skipping records, data is duplicating or missing between tools, or an integration that was working suddenly breaks. Do NOT use for Qwilr-specific automations (use /sales-qwilr-automation), general CRM platform config (use /sales-apollo or /sales-salesloft), or marketing automation flows (use /email-sequence).
This skill provides scene-by-scene analysis techniques for screenplays. Covers scene anatomy, beat breakdown, pacing evaluation, and the core question "what changes?" to identify unnecessary or weak scenes. Use when: analyzing scene effectiveness, evaluating scene pacing, identifying scene problems, or breaking down scene beats.
Turns a screenshot, mockup, Figma export, or photo of a UI into working front-end code—React + Tailwind CSS v4 by default, or Next.js, Vue, or plain HTML/CSS—matching layout, spacing, colors, and typography, then renders the result and compares it against the image to close visual gaps. Use whenever the user shares an image of a website, app screen, dashboard, component, or wireframe and wants it built, cloned, recreated, or “made real,” or says “code this up,” “build this design,” or “match this screenshot”
【Internal suite skill · Users usually don’t need to call it directly in the full workflow; simply tell the orchestrator short-drama-director what you need, and it will route to this skill automatically】【漫剧老李 V7.5.0-SLIM suite · Subskill】Specialized fight-scene storyboard engine—generates storyboard prompts for all fights, martial arts, and combat scenes: fight design (motivation / strength mismatch / stakes / environmental variables), an eight-step writing method, eight natural-sentence templates for writing each shot, strict self-checks, camera behavior locks, and a quick reference for battlefield group combat. Trigger words: fight scene, martial arts, combat storyboard, battle scene, ultimate move, sword formation, magic formation, dharma form, mowing down enemies, spell clash, moves, martial arts, boxing, sword duel, finishing move, battlefield, group battle, troop formation, camera movement lock. Suite members: orchestrator short-drama-director (authority for system locks and input format) + sd-decision-engine (decisions) + sd-script-engine (scripts) + sd-asset-engine (assets) + sd-storyboard-engine (storyboards) + sd-combat-engine (combat · this skill) + sd-feed-engine (feeding) + sd-qc-engine (quality control).
【Internal suite skill · Users usually don’t need to call it directly in the full workflow; simply tell the orchestrator short-drama-director what you need, and it will route to this skill automatically】【漫剧老李 V7.5.0-SLIM suite · Subskill】Input adaptation engine (delivery layer): the final gate from storyboard draft to prompts ready for feeding. A/B/C output modes (default C director-track clean version) + pre-output input gate (pacing/dialogue duration/word count/feedable semantics/image count/numbering consistency + material-form consistency) + material mapping and character anchoring + 《Material Patch Sheet》 + multi-segment continuity strategy + engineering-language cleanup. This package does not hold images—the user generates them on external platforms; default M1 (numbered placeholders), never request images or block due to missing images. Trigger words: convert to Seedance, generate prompts, feeding, prompt package, @imageN, material mapping, character anchoring, multi-segment continuity, over word count, clean engineering language, material form, patch sheet. This skill is part of a suite; the complete workflow starts with the orchestrator short-drama-director.
【Internal suite skill · Users usually don’t need to call it directly in the full workflow; simply tell the orchestrator short-drama-director what you need, and it will route to this skill automatically】【漫剧老李 V7.5.0-SLIM suite · Subskill · 〇·1 Source-lock for materials · Benchmark research】Research and benchmarking intelligence engine (lightweight): when the user says “I want to benchmark a hit show,” retrieve the show’s public information from across the web (including international sites) and deliver a 《Benchmark Intelligence Card》 in a fixed format. Core capabilities—short-drama source ranking list (official platform charts / data index sites / industry media / social discussions / international indexes), four research steps (set target → select sources → search → fill card), 13-field benchmark intelligence card template (including transferable and non-transferable points), and six evidence rules (no fabrication / always mark dates / always cite sources / separate firsthand information from retellings / separate facts from inferences / only the number of independent sources serves as a popularity reference). Trigger words: benchmark a hit, dig into it, research, find materials, gather information, trending topics, hit-show analysis, use this as a reference, genre research, category research, market research. This skill is part of a suite; the complete workflow starts with the orchestrator short-drama-director; the orchestrator selects the 〇·1 source lock.
Search and browse tasks on OpenAnt. Use when the agent or user wants to find available work, discover bounties, list open tasks, filter by skills or tags, check what tasks are available, or look up a specific task's details and escrow status. Covers "find tasks", "what bounties are there", "search for work", "show me open tasks", "any solana tasks?".
Web+AI security testing knowledge base. Combines 88,636 WooYun cases + Xianzhi L1-L4 methodology + GAARM 173 risks + OWASP Top 10 (LLM/ASI/WSTG). TRIGGER when the task is hands-on security testing: penetration testing, vulnerability discovery/exploitation, red team offense and defense, security auditing (SAST/DAST), CTF, AI/LLM security testing (prompt injection/jailbreak/MCP/Agent/sandbox escape). The user explicitly provides a test target (URL/code/model/Agent architecture) and the intent is "test/audit/find vulnerabilities/exploit". DO NOT trigger: - Security concept discussion ("what is XSS", "how does SQL injection work") → general Q&A - Non-security code review / debug / performance optimization → code-audit-skill or others - Fixing syntax errors / business logic bugs → general programming assistance - Pure web white-box code auditing (full project directory / Source-Sink taint analysis) → code-audit-skill - Looking up documentation by CVE number only → WebSearch - PI red team runs in Mira's built-in handbook / preview.mira.day (tier/TC-ID/resume) → mira-pi-tester - User wants to build/maintain their own markdown knowledge base/wiki ("knowledge base" here means the built-in read-only security corpus, not a user-editable wiki) → llm-wiki - Ingesting external PI/jailbreak articles into the Mira red team system (producing handbook TC cards) → mira-pi-ingest Boundary details: CTF short code snippet + exploitation idea → this Skill; full project directory + systematic white-box audit → code-audit-skill
Design security-conscious interfaces that protect users without frustrating them. Use when the user asks about MFA, password UX, breach notifications, trust indicators, secure forms, account recovery, or making security feel safe rather than scary.
The machine has to stay one machine while the camera does all the work. Lock the vehicle part by part, then fill the runtime with a numbered cut list that moves the camera every second. Use when the user wants a Seedance 2.5 / 2.0 video prompt for this kind of clip (Cars and vehicles at speed; 中文触发词: 汽车与载具速度片, 汽车, 赛车, 摩托, 载具), or asks how published cases of this type were prompted. Read references/cases.md before drafting
One person, one look, a handful of places. A head-to-toe appearance lock carries the whole clip, and each scene gets one location, one gesture and one kind of light. Use when the user wants a Seedance 2.5 / 2.0 video prompt for this kind of clip (Fashion lookbook and portrait film; 中文触发词: 时尚 lookbook 与人像写真片, 时尚, 穿搭, 人像, 写真), or asks how published cases of this type were prompted. Read references/cases.md before drafting.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.