The catalog
610,045 rows, by stacks then stars
sayed3li97/skillscore/test/fixtures/broken/legacy-notes Skill
4★ repoLint and score any AI agent SKILL.md against the official Claude, Codex, and Antigravity authoring guides — offline Dart CLI.
KimYx0207/meta-skill-creator/skills/meta-skill-creator Skill
4★ repo创建、重构或验收可复用技能包。适用于把重复工作流做成跨宿主能力包,并明确触发规则、第一动作、渐进加载、资产模板、脚本校验、触发评测、基线对比、验收证据、闭环治理、公开交付边界和不可伪造的运行证明;也适用于清理冗余参考、拆分人读模板与机器结构数据、补齐首次公开前的验收记录、记录运行反馈、生成写回或不写回决定。当用户提到"做一个 skill / 改 skill / 优化 skill / 评审 skill / 加 trigger eval / 加 release gate / 加 baseline / 准备开源 / 做闭环 / 写回改进"时,必须使用本技能;不要用普通提示词改写 SKILL.md。一旦涉及 skill 设计、改写、评审、对比、验收、量化打分、冗余清理、frontmatter 加字段、加 license、加 compatibility、闭环复盘,立刻调用本 skill。
khshanovskyi/custom-agent-skills/agent_demo/_skills/style-guide Skill
no license4★ repoEnforces brand voice, tone, and writing style rules. Use when the user asks to write, rewrite, edit, proofread, or review any text — blog posts, emails, social media copy, product descriptions, or support replies. Also use when the user asks about tone, style, or brand writing guidelines.
sandover/agent-skills/skills/claude-review Skill
no license4★ repocodex and claude skills repo
sumup/sumup-skills/skills/sumup-best-practices Skill
4★ repoPick the right SumUp integration path and apply security best practices. Use when deciding between Hosted Checkout, Card Widget, Checkouts API, mobile SDKs, terminal SDKs, or Cloud API; choosing API key vs OAuth vs restricted keys; or reviewing SumUp integration security.
composite/korean-skills/skills/egovframe-compatibility Skill
4★ repoUse when reviewing or writing eGovFrame 4.x or 5.x Java code — Controller, Service, DAO, configuration, or library usage. Covers inheritance requirements, forbidden call patterns, and architecture constraints. Assumes Spring Framework knowledge.
Elnora-AI/elnora-plugins/elnora/skills/elnora-admin Skill
4★ repoThis skill should be used when the user asks to "log in", "logout", "check auth", "create API key", "revoke API key", "list API keys", "check health", "submit feedback", "view audit log", "shell completions", "get account", "update account", "account details", "view agreements", "accept terms", "validate token", "elnora setup", "api key policy", "delete account", "list users", "feature flags", "legal documents", "set feature flag", "manage legal docs", "list profiles", "show profiles", "whoami", "run diagnostics", "open platform", or any task involving Elnora Platform authentication, administration, or diagnostics.
The same file is in 1 other repository
ntaffzii/Skill-Agents/Skill.md/cross-model-handoff Skill
4★ repoCreate continuation notes that another model, provider, or agent can resume without relying on hidden context or exact tool names. Use when the user asks to hand off work between Claude, ChatGPT, Gemini, Codex, local agents, or another MCP-enabled assistant.
cilidinezy-commits/doc-harness/kimi-skill Skill
4★ repoDocument-based project control that lets any AI agent or human resume work from files alone — no external memory needed. Use this skill whenever the user wants to structure a long-running project, track progress across sessions, recover state after context loss, coordinate multiple agents on the same project, audit project documentation health, or stop forgetting what was done last session. Triggers include: 'help me set up this project', 'I keep losing track', 'my agent forgets between sessions', 'organize my project docs', 'audit this project', 'check the documentation', 'what did we do last time', 'update the status docs', 'sync the project state', 'save everything before context compresses', 'flush context', 'prepare for compact', 'register this new file', 'init doc-harness', 'recall', 'remember', 'find in docs', 'search project docs', 'where did we discuss', 'why did we decide', 'what is the current plan', 'summarize what we know about'; multi-week projects (theses, research, analyses, software modules) that span many sessions; cross-project coordination (inbox/outbox for file-based messages between projects); requests to create CLAUDE.md, CURRENT_STATUS.md, FILE_INDEX.md, WORKLOG.md, or DOC_HARNESS_SPEC.md.
Steffd415/clawvr-agent-skill Skill
4★Generates a complete custom AI Operating System for any small business — master prompt + workflow superprompts + 30-day deployment roadmap — across 21+ verticals (dental, contractor, salon, real estate, restaurant, law, vet, photographer, fitness, insurance, CPA, and more). Use when a small business owner asks for help implementing AI in their business, when they want a custom ChatGPT/Claude prompt for their industry, or when they need a vertical-specific AI workflow.
forkadarshp/MPort Skill
4★Use when the user asks to migrate, upgrade, or port prompts, agents, tools, or codebases from one LLM model or provider version to another. Also use when auditing or continuing migration work started by another agent. Covers API calls, prompt assets, agent configs, tool definitions, tests, docs, runtime settings, rollout plans, and handoff coordination. Can also benchmark pre/post migration systems (old model and prompts vs new model and enhanced system) and report the stats as evidence.
Adri3l-R3nan/cognify-skills/.github/skills/abm-campaign-builder Skill
no license4★ repoBuilds targeted Account-Based Marketing campaigns by identifying ideal customer profiles, mapping buying committees, creating personalized outreach sequences, and designing multi-channel engagement strategies. Use when targeting enterprise accounts, building sales playbooks, or designing vertical market penetration strategies.
mateuszperlak/skill-issue Skill
no license4★This skill makes the agent completely useless. It doesn't help. It gives confidently wrong short answers and blames the user.
huybuidac/claude-code-patchkit/skills/claude-patch Skill
4★ repoPatch Claude Code CLI binary to unlock hard-coded limitations. Use when the user wants to apply, revert, or check status of binary patches after upgrading Claude Code. Community-contributed patches for model restrictions, feature flags, and schema unlocks.
Tanstack appoperation requests
hewr-srood/tanstack-appoperation-kit/.cursor/skills/tanstack-appoperation-requests Skill
no license4★ repoStandardize how API requests are defined and consumed via AppOperation together with TanStack Query. Use in React Native/Expo projects when the user mentions appOperation, QUERY_KEYS, api.ts from this kit, or wants to add new API calls, hooks, or screens that talk to the backend.
disruptionjoe/caret/research/reports/gstack-caret-it-tests/raw Skill
4★ repoDesigner's eye QA: finds visual inconsistency, spacing issues, hierarchy problems, AI slop patterns, and slow interactions — then fixes them. Iteratively fixes issues in source code, committing each fix atomically and re-verifying with before/after screenshots. For plan-mode design review (before implementation), use /plan-design-review. Use when asked to "audit the design", "visual QA", "check if it looks good", or "design polish". Proactively suggest when the user mentions visual inconsistencies or wants to polish the look of a live site.
Dragoon0x/dragoon-skills Skill
4★Generate optimal Claude Code configuration for any project. Creates CLAUDE.md, settings.json, skills, hooks, agents, and commands tailored to your stack. Run /forge to start.
The same file is in 1 other repository
efeumutaslan/SAP-SKILLS/skills/sap-abap-advanced Skill
4★ repoAdvanced ABAP development skill extending the base sap-abap skill. Use when working with ABAP Cloud (Tier 1) vs Classic (Tier 2), AMDP for HANA pushdown, abapGit workflows, released API checks, or RTTI/RTTC dynamic programming. If the user mentions ABAP Cloud, Steampunk, AMDP, abapGit, Tier 1/Tier 2 API, or classic-to-cloud migration, use this skill.
0xlayerghost/solidity-agent-kit/skills/claude-code-usage Skill
4★ repo[AUTO-INVOKE] MUST be invoked at the START of each new coding session. Covers context management, task strategies, and Foundry-specific workflows. Trigger: beginning of any new conversation or coding session in a Solidity/Foundry project.
The same file is in 1 other repository
Laz1mov/L1-AGENT-WALLET/skills/bitcoin-governance Skill
no license4★ repoUse this skill when the user asks about wallet policy, spending limits, governance, allowance, or wants to change how much the agent can spend autonomously. Triggers on: "what is my allowance", "change my spending limit", "update policy", "how much can you spend", "governance", "policy version", "whale threshold", "recovery path", or any request to inspect or modify the Taproot governance structure.
DaleStudy/skills/skills/bun Skill
4★ repoNode.js 대신 Bun 런타임 사용을 위한 스킬. 다음 상황에서 사용: (1) 새 JavaScript/TypeScript 프로젝트 생성 시, (2) package.json 또는 의존성 관련 작업 시, (3) 스크립트 실행 또는 개발 서버 시작 시, (4) Node.js 프로젝트를 Bun으로 전환 시, (5) 'bun', 'bunx', 'bun.lockb' 키워드가 포함된 작업 시
The same file is in 1 other repository
wicttor/pwrl/pwrl-end-session-checkpoint Skill
no license4★ repoVerify repository state and confirm session completion before committing.
Sagargupta16/skillcheck/fixtures/broken-skill Skill
4★ repoConformance suite for Agent Skills: lint SKILL.md against the spec, run against real agent runtimes, publish a compatibility matrix
netresearch/typo3-testing-skill/skills/typo3-testing Skill
no license4★ repoUse when setting up TYPO3 extension test infrastructure, writing unit/functional/E2E tests, configuring PHPUnit 11/12/13, mutation testing, mocking final classes (v14), CI/CD matrix across TYPO3 12/13/14.3 LTS, dev-dependency consolidation via typo3-ci-workflows meta-package, or debugging CI failures. Also triggers on: testing-framework setup, ensure proper testing, test matrix, integration testing, e2e testing, coverage, test generation.
The same file is in 1 other repository
zerocracy/zealot/skills/file-bug-report Skill
4★ repoUse this skill when the user wants to file a single bug as a new GitHub issue against a specific repository.
Architect-SIS/sis-skill Skill
4★Sovereign Intelligence System - Equilibrium-native reasoning for OpenClaw Safety
hardness1020/awesome-prompt-skill/skills/prompt-engineering Skill
4★ repoExpert guidance for designing, optimizing, evaluating, and securing prompts and system prompt architectures for LLMs. Use when users need help with writing or improving prompts, designing system prompts or multi-section prompt architectures, building agent prompts with tool integration, prompt optimization and automated tuning, prompt security and injection defense, prompt evaluation and benchmarking, production prompt management, or understanding prompt engineering techniques like Chain of Thought, ReAct, Tree of Thoughts, few-shot learning, and Constitutional AI. Covers patterns derived from production agentic systems and the broader prompt engineering research landscape.
Ampli-Group/agentic-mobile-blueprint/.agents/skills/apple-appstore-setup Skill
4★ repoSet up Apple Developer account, App Store Connect app listing, and EAS credentials for iOS builds and TestFlight distribution. Use when creating a new iOS app, configuring Apple certificates/provisioning, or troubleshooting App Store Connect submission issues.
agoradynamics/wick/.claude/skills/wick-automate Skill
no license4★ repoDetect when a task is being repeated often enough to be worth automating, then propose the right automation — a PROGRAM (deterministic script) or a SKILL (reusable judgment procedure). Analyzes the current session, the memory/ record, or an observation log; classifies program-vs-skill; estimates payoff; drafts the automation. Proposes, never auto-builds. Use when you catch yourself or the user repeating a multi-step task, or to mine an observation log periodically.
acnlabs/anyone-skill Skill
no license4★Distill anyone into a runnable OpenPersona skill pack — real or fictional, personal or public, living or historical. Collects chat logs, documents, and public content, extracts a 4-dimension persona, and generates a portable OpenPersona pack via skills/open-persona. Use when asked to distill, clone, or create a persona for any person or character.
CharksChen/oh-my-teacher Skill
no license4★面向中文高校场景的期末复习助教:课程画像、资料整理、重点提炼、往年题分析、 自适应刷题、严格批改(踩点给分+多维认知诊断)、错题本(误区修复系统)、 费曼讲解、苏格拉底追问、角色逆转互教、主动回忆、间隔复习(连续重学+知识类型差异化)、 可视化讲解、代码演示、考前冲刺、考试就绪度评估、考试心理与临场策略、 信心校准、级联失效追溯和复盘闭环。Use when the user asks to study or review a university course, organize PDFs/PPTs/notes/homework/past papers, prepare for paper/lab/oral/coding exams, generate quizzes/flashcards/mock exams, grade answers, fix weak points, build a cram plan, enable opt-in reminders, request daily/weekly knowledge digests, assess exam readiness, practice exam strategies, do peer-quiz role reversal, calibrate confidence, or says 期末复习, 课程材料, 课件, 老师划重点, 出题, 批改, 错题, 背诵, 苏格拉底, 费曼, 考前冲刺, 闭卷, 开卷, 机考, 实验考, 往年题, 题库, 知识库, 笔记, 课程主页, 错题本, 今天该复习什么, 考前速记 PPT, 复习仪表盘, 考试就绪度, 角色互换, 信心校准, 考试策略, 卡题恢复, ima, NotebookLM, Obsidian.
Cpp1022/concise Skill
4★Chinese-first concise mode SKILL.md for Claude Code / Codex CLI / Cursor — shorter, denser replies without losing technical accuracy.
wzszsw/easy-query-orm Skill
no license4★Use when Java/Kotlin code uses easy-query (often abbreviated as `eq`; `EasyEntityQuery`, `EasyQueryClient`, proxy DSL) or when migrating JPA/MyBatis to easy-query. Covers setup and missing `*Proxy` troubleshooting, Spring Boot starter/config, CRUD and mutation APIs (`insertable/updatable/deletable`, `onConflictThen`, `mapInsertable/mapUpdatable`, `setSQLStrategy`, version/logic delete), diff update tracking, search-form DTO query (`whereObject`, `orderByObject`, `EasySearch`), relation metadata/loading (`@Navigate`, `include`, `selectAutoInclude`), implicit and explicit subquery/group-join/CTE/window/UNION analytics, `savable`, sharding, code-first DDL, and verified API lookup/troubleshooting. Do not use for unrelated ORMs.
yarlson/yarstack/plugins/yarstack/skills/alternatives-explore Skill
4★ repoGenerate and evaluate genuinely different approaches to a product, workflow, or system problem. Use when the user explicitly wants a bold, contrarian, experimental, or non-incremental alternative and a testable recommendation.
Wang-Cankun/skillshelf/app/src/lib/data/skills/sample-chatroom Skill
4★ repoA demo multi-persona discussion skill. Three role-played experts — an optimist, a skeptic, and a moderator — debate a question and converge on a recommendation. Triggers: /sample-chatroom, "panel discussion", "debate this".
sametcelikbicak/task-decomposer Skill
4★Use when the user has a large, complex, or ambiguous request that needs to be broken down into smaller, actionable steps for AI assistance. Helps users structure prompts, manage context window limits, plan multi-session work, and get better results from AI coding agents. Also for 'too complex', 'where do I start', 'break this down', 'step by step', 'context too large', 'multi-step task', 'prompt structure', 'AI workflow', 'organize my request'.
frankxai/awesome-hermes-agent-skills/skills/coding-agents-superpack Skill
no license4★ repoUse when orchestrating multiple coding agent CLIs (Claude Code, Codex, OpenCode, Grok CLI, Antigravity). Discovery, comparison, structured prompting, and multi-agent handoff patterns. Portable free pack — no host-specific secrets.
webkubor/keyring Skill
4★轻量级加密密钥管理 — 支持多账户多密钥,AI 安全注入。触发条件: 需要管理密钥、密码、API Token 时触发。触发词: 密钥、secret、token、API key、password、keyring。
AltrinaAI/vibestudio/examples/docx Skill
4★ repoUse this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation.
The same file is in 31 other repositories
ThanhNguyxnOrg/blendops/bundles/claude-desktop-manual Skill
4★ repoClaude Desktop-facing BlendOps router skill for planning, runtime readiness, evidence truth, and non-Blender-user outputs using bundled references.
tonbistudio/gitlana/examples/onchain-skill-resolver Skill
4★ repoFetch, verify, and install agent skills stored entirely on the Solana blockchain, given only a Metaplex Core asset address.
tbc-servicos/dataagile-agent-kit/fluig/skills/api-ref Skill
4★ repoReferência das APIs TOTVS Fluig — DatasetFactory, DatasetBuilder, CardAPI, WCMAPI, fluigc, WFMovementDTO, getValue/setValue. Use quando tiver dúvida sobre qual API usar no Fluig, sintaxe de dataset, como manipular formulário, ou diferença entre APIs de workflow e formulário.
langgenius/mosoo-connector/publish/skills/mosoo Skill
no license4★ repoUse when a coding agent needs to work with Mosoo setup, local or cloud runtime state, Mosoo CLI operations, or app integration with a published Mosoo Agent.
osmedeus/osmedeus-skills/osmedeus-expert Skill
no license4★ repoExpert guide for the Osmedeus security automation workflow engine. Use when: (1) writing or editing YAML workflows (modules and flows), (2) running osmedeus CLI commands (scan, workflow management, installation, server), (3) configuring steps, runners, triggers, or template variables, (4) debugging workflow execution issues, (5) building security scanning pipelines, (6) working with agent/LLM step types, or (7) any question about osmedeus features, architecture, or best practices.
musabekisakov-imj/flutter-design-engineer/skills/flutter-accessibility Skill
4★ repoAudit and improve Flutter accessibility, inclusive interaction, localization safety, and RTL behavior. Use for screen-reader semantics, focus and keyboard navigation, text scaling, contrast, touch targets, reduced motion, form errors, RTL, localization, or accessibility acceptance checks.
yasserstudio/perfex-crm-skills/skills/perfex-core-apis Skill
4★ repoUse whenever the user is working inside a Perfex CRM codebase and touches `get_option`, `update_option`, `add_option`, `delete_option`, `hooks()`, `do_action`, `apply_filters`, `register_activation_hook`, `$this->load`, `get_instance()`, `$CI`, `db_prefix()`, auth helpers like `is_staff_logged_in` / `get_staff_user_id` / `staff_can`, or `_l()`. Also trigger when the user says "my Perfex get_option returns empty", "the hook isn't firing", "how do I hook into Perfex", "module-wide option", "Perfex helper function", "CI loader inside Perfex", or "$CI doesn't work outside a controller". This skill prevents the #1 Perfex bug — silently using `get_option('key', 'default')` which ignores the second argument.
stephenrogan/augmented-csm/skills/pillar-1-book-intelligence/bi-account-brief Skill
4★ repoCompiles a comprehensive structured account snapshot by pulling from all available data sources. Gives the CSM full account context in under 60 seconds instead of 30-60 minutes of manual assembly. Use when asked to prepare for a customer call, build an account brief, generate a QBR prep summary, create an account snapshot, review an account before a meeting, assemble account context for an escalation, or when any human decision point across any pillar needs current account context. Also triggers for pre-call prep, account review, executive briefing preparation, or any request for a consolidated view of an account.
keshavjoglekar/auditing-agent-config Skill
4★Audits a Claude Code and/or Claude.ai configuration for redundancy, misplacement, and conflicts — across global (~/.claude) and project (.claude/) scopes, skills, plugins, CLAUDE.md/rules, subagents, hooks, and MCP/connector setup. Use this whenever the user wants to audit, review, clean up, reorganize, or sanity-check their Claude setup; mentions having "too many skills/plugins"; asks whether something should be global or project-level; suspects duplicate, conflicting, or overlapping skills; wants to know if their CLAUDE.md is too long or bloated; asks "why isn't skill X triggering"; or asks things like "audit my claude setup," "check my skills for overlaps," or "help me organize my Claude Code config." Trigger proactively any time the user is actively managing a nontrivial skill/plugin library (roughly 10+ skills or 3+ plugins) and starts talking about adding more.
yfge/skill-finder Skill
4★Search, discover, and install OpenClaw agent skills from multiple registries. Use when: user asks "find a skill for X", "is there a skill that can...", "how do I do X" (where X might have an existing skill), wants to browse available skills, or asks to install/add a skill. Also triggers on "search skills", "explore skills", "what skills exist for...", or "extend capabilities".
huxleyli15/frank/frank Skill
4★ repoKills sycophancy. Stops the assistant from flattering, hedging, and caving. No "great question", no "you're absolutely right", no folding the moment you push back. Leads with the verdict and the risk, steelmans your idea then attacks it, labels every claim by confidence, and holds its ground on evidence instead of tone. Praise only when it's specific and earned. Use on ANY request where you want an honest second opinion instead of a cheerleader: reviewing a plan, a design, an argument, code, writing, or a decision. Also activate whenever the user says "frank", "be honest", "be blunt", "no flattery", "don't agree just to agree", "push back", "steelman", "诤友", "说实话", "别拍马屁", "别附和", "反驳我", or "唱反调". Do NOT manufacture disagreement for its own sake, and do NOT be rude — attack the idea, not the person.
n-shadloo/secure-code-auditor Skill
4★Backend security auditor with deep Django and Django REST Framework coverage layered on a general OWASP Top 10 (2025) and API Security Top 10 (2023) foundation. Use whenever backend code is being written or reviewed and security is in scope — including any time the work touches authentication, login, sessions, JWT, OAuth2/OIDC, social login, API keys or tokens, permissions or access control, user-supplied input, the ORM or raw SQL, file uploads, serializers or API endpoints, secrets or settings, payments, email or notifications, background tasks, caching, async/ASGI or WebSockets, signals or lifecycle hooks, migrations, or deployment configuration, even if the word "security" is never used. Runs in two modes: review-time (audit existing code and return prioritized, actionable findings with severity, location, and a concrete fix) and write-time (apply secure defaults and flag risky patterns while generating code). Django/DRF is the primary target; the general layer makes it useful for any backend stack.
celigo/ai/skills/building-apis Skill
4★ repoDomain knowledge and tools for building Celigo integrations with AI coding assistants.
entr0pia/scoop-skills/skills/discover-shims Skill
4★ repoDiscover available Scoop shims and identify valuable command-line tools. Use when the user asks about available tools, what commands they can use, or wants to explore their Scoop environment.
nwyin/labrat/labrat Skill
4★ repoRun autonomous ML research sessions: reproduce paper results, test hypotheses, run ablations — all within a compute budget. Use when the user wants to run ML experiments on Modal, mentions 'labrat', provides a research plan or experiment spec, or asks to run overnight experiments. Also use when continuing an existing research session (`.research/state.json` exists).
gabrieldabbah/genesis/skills/axiomatic-induction Skill
4★ repoUltimate Opus 5 builder. Creates and transforms any setup into the ultimate setup for generation 5 models.
Agent skill maccabi pharm search
alexpolonsky/agent-skill-maccabi-pharm-search Skill
4★Search for medications and check real-time stock availability at Maccabi pharmacies in Israel. Use when searching for drugs like "nurofen", "acamol/אקמול", "advil", or finding nearby pharmacy branches with stock. Supports Hebrew and English drug names. מכבי פארם, תרופות, מלאי, בית מרקחת.
kt3k/skills/skills/breakdown Skill
no license4★ repoBreak down a given task into subtasks with time estimates. Analyzes the codebase and task requirements to produce a structured plan with estimated effort for each step.
scasella/codex-gpt-pro/gpt-pro Skill
4★ repoOperate ChatGPT Pro through Browser Use/Playwright or Computer Use. Use when the user asks Codex to ask GPT Pro, prompt GPT Pro, use ChatGPT Pro, send a prompt to ChatGPT Pro/GPT Pro, or wait for a GPT Pro result. Checks whether Browser Use/Playwright and/or Computer Use are available, asks for a first-run default route when both are available, selects Pro, submits prompts, waits through long reasoning or streaming, and returns the completed answer while pausing for login, MFA, CAPTCHA, account-security prompts, and never exporting or saving cookies, session tokens, or auth credentials.
gtskevin/survey-scale-review Skill
4★Use when reviewing research survey scales, for organizational behavior, management, marketing, psychology, and social science research questionnaires with English source items, Chinese translations, adapted scales, reverse-coded items, paired leader-employee surveys, multi-wave surveys, or pre-launch questionnaire quality checks.
Mr-DS-ML-85/searxng-websearch Skill
4★Use this skill whenever the user wants to search the web, do research on a topic, fetch a webpage, or gather information from online sources using a self-hosted SearXNG instance. Triggers include: 'search for', 'look up', 'find online', 'deep research', 'fetch this page', 'web search', or any request that needs current/external information. Also use when the user wants to run websearch.py, deep_research.py, or fetch_page.py scripts. Do NOT use for local file operations, code execution unrelated to search, or tasks that don't require web data.