Skills
101,323 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Complete FFmpeg video stabilization and 360/VR video processing. PROACTIVELY activate for: (1) Video stabilization (deshake, vidstab), (2) Hardware-accelerated stabilization (deshake_opencl), (3) 360/VR video transforms (v360), (4) Perspective correction (perspective), (5) Ken Burns/zoom-pan effects (zoompan), (6) Lens distortion correction (lenscorrection, lensfun), (7) Action camera footage, (8) Drone video processing, (9) VR headset formats. Provides: Stabilization workflows, 360 projection conversions, motion effects, lens correction.
Provides domain knowledge and guidance for the Flare Time Series Oracle (FTSO)—block-latency feeds, Scaling anchor feeds, feed IDs, onchain and offchain consumption, fee calculation, delegation, and smart contract integration. Use when working with FTSO, price feeds, oracle data, feed consumption, volatility incentives, or Flare Developer Hub FTSO guides and starter repos.
Edit images with Flux 1 Kontext Pro (Black Forest Labs' precise local image-edit model) on RunComfy — bundled with the model's documented prompting patterns so the skill gets sharper output than naive prompting against the same model. Documents Flux Kontext's strengths (single-reference precise local edits, strong prompt control, consistent high-fidelity outputs), the schema (single image + prompt), and when to route to Nano Banana Edit / GPT Image 2 edit / Flux 2 Klein instead. Calls `runcomfy run blackforestlabs/flux-1-kontext/pro/edit` through the local RunComfy CLI. Triggers on "flux kontext", "flux-kontext", "flux 1 kontext", "kontext", "BFL kontext", or any explicit ask to edit with this model
Use when debugging or handling API errors in Frappe/ERPNext v14/v15/v16. Prevents silent failures and wrong HTTP status codes in REST endpoints. Covers 401 Unauthorized (wrong token format, expired OAuth), 403 Forbidden (missing @whitelist, allow_guest needed), 404 Not Found (wrong endpoint URL), 417 Expectation Failed (validation via frappe.throw), 500 Internal Server Error, CORS issues, CSRF token missing/invalid, rate limit exceeded (429), file upload failures, JSON parse errors in request/response, webhook delivery failures, and timeout on long operations. Keywords: API error, 401, 403, 404, 417, 429, 500, CSRF, CORS, REST, API call fails, 403 forbidden, CORS error, token expired, endpoint not found, webhook not received. whitelist, webhook, rate limit, file upload, authentication token.
Use when debugging hooks.py errors in Frappe/ERPNext. Covers hook not firing (typo, wrong dict structure), circular imports, app_include_js path errors, scheduler_events not running, doc_events on wrong DocType, permission_query_conditions SQL errors, override_doctype_class import failures, extend_doctype_class [v16+] conflicts, fixtures not loading. Error diagnosis by hook type for v14/v15/v16. Keywords: hooks.py error, hook not firing, scheduler not running,, hook not working, scheduler not running, app_include not loading, override not applied. doc_events error, circular import, fixtures error, override class error.
Use when building Jinja templates in Frappe: Print Formats, Email Templates, Notification templates, Portal Pages, and custom Jinja methods. Covers template creation workflows, child table handling, conditional sections, styling, multi-language support, and debugging. Prevents N+1 queries, wrong formatting, and Report Print confusion. Keywords: create print format, email template, portal page, pdf, create print format, invoice template, email template, PDF layout, custom print. template, invoice template, jinja methods, notification template, web page template, print format styling.
Use when deploying Frappe/ERPNext to production, configuring Nginx or Supervisor, setting up Docker, enabling SSL, or hardening security. Prevents insecure deployments, missing reverse proxy config, and broken process management. Covers production setup, Nginx configuration, Supervisor/systemd, Docker Compose, Let's Encrypt SSL, firewall rules, security hardening. Keywords: deployment, production, nginx, supervisor, docker, ssl, letsencrypt, security, gunicorn, systemd, go live, production setup, HTTPS setup, server config, deploy to VPS, Docker setup..
Use when building database queries with frappe.qb in Frappe v14-v16. Covers the PyPika-based query builder: SELECT, INSERT, UPDATE, DELETE, joins, aggregation, subqueries, cross-DB compatibility (MariaDB/PostgreSQL), and migration from raw SQL. Prevents SQL injection and DB-specific bugs. Keywords: frappe.qb, query builder, DocType, Field, pypika, join, frappe.qb example, how to write query, join tables, SQL replacement, parameterized query, aggregate, ImportMapper, cross-database, parameterized query.
Use when setting up CI/CD pipelines for Frappe apps, configuring GitHub Actions test workflows, or adding linting and security scanning. Prevents broken CI from incorrect test matrix configuration, missing MariaDB/Redis services, and uncaught code quality issues. Covers GitHub Actions workflows, test matrix (Python/Node versions), semgrep rules, pre-commit hooks, linting (ruff, eslint), CI test environment setup. Keywords: CI/CD, GitHub Actions, test matrix, semgrep, pre-commit, linting, ruff, eslint, continuous integration, automated tests, GitHub Actions, CI pipeline, pre-commit, code quality check..
Benchmark a third-party coding-agent harness against FrontierHarness Eval using Runta runtimes. Provisions a clean runtime bound to the GitHub repo under evaluation, installs the Harbor and Pier stacks needed for Terminal-Bench and DeepSWE tasks, freezes a golden checkpoint, runs tasks from identical fresh restores while saving trajectories as evidence, generates a comparison diagram, and builds a shareable report. Use when evaluating, benchmarking, scoring, or comparing a coding agent harness, or when the user mentions FrontierHarness, DeepSWE, Terminal-Bench, Harbor, Pier, golden checkpoints, or harness trajectories.
Configure GrowthBook API credentials and the trusted app origin so the other skills can run and link to the UI. Use when the user says "set up growthbook", "configure my api key", "growthbook isn't working", "where do I put my key", or when another skill emits an error pointing here ("GB_API_KEY is not set", "authentication failed", "GB_APP_URL is required"). Writes ~/.config/growthbook/.env with chmod 600 and validates against the live API. For listing flags or running experiments, the domain skills handle that.
Use when writing German-language prose longer than one sentence for Jira tickets, internal German wiki/spec docs, or team-chat to German-speaking colleagues (Slack, Matrix, Teams). Invoke BEFORE composing. Covers anglicism verbs (failt, triggern, returnen), calqued collocations and idioms (blockiert auf, macht Sinn, in 2026), loanword genders (der Commit, die Pipeline), per-artifact register (impersonal for Jira/wiki, ich/du for chat), German typography (Gedankenstrich, Anführungszeichen, Durchkopplung), and AI-typical rhythm (restatement, symmetric paragraphs, summary closers). Skip for commit messages, MR/PR descriptions, release notes (English by team convention), conversational chat replies, or single-line acknowledgments. If you catch yourself thinking 'my German is probably fine here' for anything longer than a sentence — invoke this skill.
Salesforce Metadata API reference for creating, understanding, or modifying metadata XML files (.object-meta.xml, .flow-meta.xml, etc.) and any of 604 metadata types (CustomObject, Flow, ApexClass, ApexTrigger, Profile, PermissionSet, Layout, ValidationRule, RecordType, EmailTemplate, ...). Use for questions about a type's fields or XML format, or Salesforce DX project structure (force-app/main/default). TRIGGER when: authoring or editing any *-meta.xml file, asking what fields/format a metadata type has, or whenever 'Salesforce metadata' or 'sfdx project' is mentioned. DO NOT TRIGGER when: SOQL/DML/runtime sObject access (use the Enterprise API skill), Tooling API developer records (ApexExecutionOverlayAction, TraceFlag), or non-XML logic (Apex code, LWC JS, Visualforce controllers).
[AUTO-INVOKE] MUST be invoked BEFORE creating git commits, PRs, or code reviews. Covers Conventional Commits, PR templates, review requirements, and AI-assisted development rules. Trigger: any task involving git commit, git push, PR creation, or code review.
Create GitLab project or personal code snippets with glab. Use when sharing code or text as a new snippet from files or stdin. For viewing, editing, or deleting existing snippets, use the GitLab UI or the snippets API through glab api. Triggers on snippet, gist, code snippet, share code, create snippet.
Manage your GitLab to-do list from the CLI, including listing pending or completed items and marking items done. Use when reviewing assigned or mentioned work from GitLab notifications, clearing to-dos, or scripting to-do triage. Triggers on todo, to-do, notification list, mark done, assigned items, mentioned items.
Expert Go developer specializing in Go 1.21+ features, concurrent programming with goroutines and channels, and comprehensive stdlib utilization. This agent excels at building high-performance, concurrent systems with idiomatic Go patterns and robust error handling.
GPT Image workflow assistant for text-to-image, reference-image editing, inpainting instructions, posters, typography, UI mockups, diagrams, product images, and final image prompt preparation. Use when the user explicitly wants GPT Image, GPT Image 2, gpt-image, image generation, or image editing. This skill does not include scripts, reference files, or automatic setup.
Edit images with OpenAI GPT Image 2 (the `/edit` endpoint of ChatGPT Images 2.0) on RunComfy — bundled with the model's documented prompting patterns so the skill gets sharper output than naive prompting against the same model. Documents GPT Image Edit's strengths (preservation language, multilingual in-image text editing, multi-reference up to 10 images, layout / typography precision), the schema, and when to route to Nano Banana Edit / Flux Kontext / GPT Image 2 t2i instead. Calls `runcomfy run openai/gpt-image-2/edit` through the local RunComfy CLI. Triggers on "gpt image edit", "gpt-image-edit", "chatgpt image edit", "edit with gpt image 2", or any explicit ask to edit with this model
Interview the user relentlessly about a plan or design until every branch of the decision tree has an explicit recorded decision and no open questions remain. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Browse and query the documentation for Hexagone frontend components (@his/hexa-components). Use when the user asks about Hexagone Vue.js components, UI patterns, beta-scss CSS classes, Vuex store modules, custom directives, form validation rules, or the Hexagone frontend design system. Fetches LLM-optimized documentation from the GitLab repository.
Use when users need to diagnose I2C bring-up failures on Arduino or ESP32-class boards, including no devices found, wrong readings, address conflicts, scanner-detects-but-library-fails cases, pull-up problems, or board-family-specific bus quirks.
Draft, analyse, score, and improve IELTS Academic Writing Task 1 responses (150-word minimum reports on graphs, charts, tables, diagrams, maps, or processes). Use this skill when the user shares a Task 1 prompt, wants a model answer, needs band-score feedback on their own response, or asks how to describe trends, compare data, or structure an overview. Also use when the user asks about Task 1 vocabulary, grammar for describing data, or common Task 1 mistakes. Covers all visual types: line graph, bar chart, pie chart, table, map, process diagram, multiple charts.
Mask-driven image inpainting on RunComfy via the `runcomfy` CLI. Routes to Tongyi MAI Z-Image Turbo Inpainting (the dedicated inpainting endpoint with mask, strength, and control-scale) and to identity-preserving edit models (Nano Banana 2 Edit, GPT Image 2 Edit, FLUX Kontext Pro) when a mask isn't available and the region must be described instead. Use for object removal, watermark removal, region replacement, blemish cleanup, and any controlled local edit where a binary mask defines the target area. Triggers on "inpaint", "inpainting", "image inpaint", "remove from image", "fill region", "mask-driven edit", "remove watermark", "remove object", "patch the photo", "fill the hole", or any explicit ask to edit a specific masked region of a still.
A profile-driven coaching assistant for building an IP. On first use, create a private profile; afterwards, combine the profile, historical data and current evidence to re-diagnose the real problem, then work on persona and account positioning, content pillars, topic sourcing, evaluation and revision, short-video scripts and voiceover copy, on-camera performance intent, account launch and follower growth, series and account memory, content retrospectives, or content monetization. Also used for entry requests such as "新手入门", "怎么用" and "更新 ip-strategist". Natural language or the host-supported /ip-strategist both route through the same entry point.
Build JUCE audio plugins (VST3, AU, AAX, Standalone, LV2) in C++ with CMake. Covers the full development workflow from project scaffolding to multi-format builds, including AudioProcessor lifecycle, parameter management with APVTS, DSP module chains, custom editor GUIs, Three.js/GLB WebView interfaces, state serialization, real-time audio safety, and cross-platform CI/CD. Use this skill whenever the user mentions JUCE, audio plugins, VST plugins, AU plugins, audio effects, synthesizers, MIDI processors, AudioProcessor, AudioProcessorEditor, Projucer, or wants to create audio software — even if they just say "make a plugin" or "audio plugin". Also applies when users are working in an existing JUCE codebase (files like PluginProcessor.cpp, PluginEditor.cpp, CMakeLists.txt with juce_add_plugin), or investigating a plugin's CPU usage in a host, porting a plugin to Windows or Linux, or code signing, notarizing or packaging plugin releases.
Creates a Check for Understanding (CFU) for a math topic or standard: 1–3 targeted formative items whose distractors come from documented student misconceptions, plus a teacher guide routing each likely response to a specific next instructional step. Load BEFORE asking any clarifying question about grade, standard, or topic. This skill is for math; for other subjects, help as you normally would and don't mention the skill to the user. Triggers on asks to create, write, generate, or design a check for understanding, formative check, quick check, or exit ticket for a math concept or standard, and on implicit signals like "I need to see who got it before I move on." Not a quiz, test, or practice set — a CFU is read for what it reveals, not scored, so a numbered quiz or a bare answer key is a different job. Does not create a lesson (k12-lesson-plan-creation, which brings its own exit ticket), tier one (k12-lesson-differentiation), or prepare a teacher to teach one (k12-lesson-prep).
Expert Kubernetes Specialist with deep expertise in container orchestration, cluster management, and cloud-native applications. Proficient in Kubernetes architecture, Helm charts, operators, and multi-cluster management across EKS, AKS, GKE, and on-premises deployments.
Trigger this skill when a structured summary is needed for legal documents such as judgments, rulings, mediation agreements, arbitration awards, and administrative penalty decisions. Typical triggers include but are not limited to: - the user asks for a summary of one legal document, key point extraction, or a generated holding - the user needs to turn multiple legal documents into standardized, comparable, searchable summaries - the user asks to extract case information, points of dispute, legal basis, or reasoning - the user needs summary material for case studies, litigation analysis, or lesson preparation - the user asks to distinguish the summary focus of different document types (judgment/ruling/mediation/arbitration/administrative penalty). The core value of this skill: on the basis of fidelity to the source text, turn lengthy legal documents into concise, objective, structured summaries, highlight the core content, and avoid restating the full text and subjective evaluation.
Triggered when legally significant facts need to be extracted from unstructured text such as case descriptions, party statements, chat logs, or media reports. Typical trigger scenarios include but are not limited to: - A user provides a case description in everyday language and asks for legal facts to be extracted - A user asks to convert a colloquial, emotional narrative into a fact structure usable for legal analysis - A user asks to distinguish "objective facts", "subjective judgments", and "legal evaluations" in a case narrative - A user asks to map extracted facts to the constituent elements of a specific legal relationship - A user asks to organize a case timeline or produce a fact-summary table - A user asks to convert everyday language into legal language (e.g. "he kept putting off repaying" → "failed to perform after the performance period expired") Core value of this skill: organizing mixed everyday narratives into a fact structure that can enter legal reasoning — a prerequisite step for legal analysis.
Scrape LinkedIn profiles using the user's Chrome profile. Use when asked to find leads, scrape LinkedIn profiles, extract contact data from LinkedIn, or build prospect lists. Triggers include "find founders on LinkedIn", "scrape this LinkedIn profile", "get LinkedIn data for these people", "build a lead list from LinkedIn".
Amazon reverse product selection: an indicator data pool compiled from historical business insights reports. Reverse-screen Amazon niches and keywords across 30+ business dimensions, including market size and growth, price ranges and tier shares, competition density and top-player concentration, audience profiles such as age, gender, and income, and review selling points and pain points. Trigger this skill when users mention reverse product selection, metric-based filtering, niche-market lookup, blue-ocean niche discovery, low-competition niches, beginner-friendly niches, fragmented-brand markets, pain-point entry points, selling-point lookup, pricing-tier opportunities, audience-profile-based product selection, Amazon niche reverse search, niche metrics filter, low-competition niche, blue ocean niche, demographic-based selection, pain-point niche, price tier opportunity, sweet spot pricing, or brand fragmentation. Even if the user does not explicitly say “reverse product selection,” trigger this skill whenever they want to filter Amazon niches based on business criteria.
Loads documents fully into the main agent's context so the agent can answer questions, summarize, or work with that content in subsequent turns. Use whenever the user wants to ingest, read, absorb, or pull in documents — especially when they say things like "load these docs", "read all of these", "ingest this folder", "pull in these PDFs", "load all docs in X", or paste a list of file paths/URLs and ask you to read them. Handles local files (text, code, markdown, PDFs, notebooks, images), entire folders (recursively), and remote URLs. To convert files into Markdown, use to-markdown instead.
Design well-structured agent loops with best-practice coaching and cross-model review gates before you run them. Use when the user wants to design, build, or set up an agent loop, iterative agent workflow, self-review loop, LLM-as-judge loop, multi-model council, reviewer/judge gate, or goal-driven looping process. Guides goal refinement, typed verification criteria, reviewer/judge selection, privacy boundaries, termination guards, and observability, then emits a RUN_IN_SESSION.md handoff prompt plus portable loop.yaml, loop.resolved.json, LOOP.md, and run-loop.py.
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.
