Skills
100,928 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Draft anonymous one-page company teasers for A-share sell-side M&A processes. Creates compelling blind profiles without revealing the company's identity, adapted for Chinese market conventions. Triggers on "A股Teaser", "盲审材料", "blind profile China", "anonymous teaser A-share", or "sell-side teaser".
Post-close value creation plans for China-focused portfolio companies. Adapted from the original value-creation-plan skill for Chinese business context, management practices, and market dynamics. Triggers on "A股投后增值", "价值提升计划", "value creation plan China", "100-day plan China", "portfolio value creation", or "value creation [company]".
Create professional Excel workbooks for A-share financial analysis. Adapts the original xlsx-author skill for Chinese financial modeling standards, CAS conventions, and A-share formatting. Triggers on "A股Excel模型", "财务模型制作", "create model China", "build model xlsx", "制作模型", or "Excel model [company]".
Provides expertise on Chroma Cloud integration for semantic search and hybrid search applications. Use when the user is working with Chroma Cloud, CloudClient, managed collections, Schema(), Search(), hybrid search, or Chroma Cloud CLI workflows.
Build a data-backed client onboarding proposal for an agency starting a new engagement, combining an SE Ranking SEO + AI-search audit with a Planable analysis of the client's current social performance. Use this skill whenever an agency wants to show a new (or about-to-sign) client where they stand and what the plan is, or says things like "build an onboarding proposal for [client]", "we're starting with [client], put together the SEO + social analysis", "create a client proposal with audit and social baseline", "scope a SEO + AI + social engagement", or "show the client their opportunities and our plan". Assumes the client's social accounts are already connected in Planable. Always activate for new-engagement proposals that pair SE Ranking analysis with Planable social data.
Generates fail-closed draft JSON templates and runs local deterministic structure and consistency checks for clinical reports: CARE case reports, radiology, pathology and lab scaffolds, CONSORT 2025, SPIRIT 2025, ICH E3 CSRs, aggregate adverse-event tables, provenance manifests, terminology schemas, and de-identification process checklists. Use when drafting a case report for publication from a verified source-fact manifest. Use when checking a trial results manuscript, protocol, or CSR draft against its reporting structure. Use when building an aggregate safety table from synthetic or de-identified data. Use when routing an artifact to the right guideline. Not for diagnosis, treatment advice, individual case safety report creation, or filing and submission.
Runs constraint-based metabolic modeling with COBRApy (Python, import cobra) on genome-scale models in SBML, JSON, YAML, or MATLAB format. Covers FBA, pFBA, geometric FBA, FVA, flux sampling, gene and reaction knockouts, production envelopes, growth media, gapfilling, and building models. Use when loading or exporting a genome-scale metabolic model. Use when predicting growth or flux distributions with FBA or FVA. Use when screening gene or reaction knockouts. Use when tuning growth media or exchange constraints. Use when gap-filling an infeasible model or checking model consistency. Not for kinetic or ODE-based simulation of metabolism.
Use when the user asks to improve, add, or clean up code comments, focusing on WHY over WHAT. Reads specified files (or recently modified ones), adds strategic inline comments for non-obvious logic, removes redundant ones, and converts WHAT-comments to WHY-comments. Accepts optional $ARGUMENTS for target files.
Audit a codebase for maintenance and modernization. Challenges scope, reviews architecture/quality/tests/performance/dependencies, files deferred work via bd. Language-specific addendums for iOS/Swift, Go, and Web/JS/CSS activate automatically based on what's in the repo. Supports monorepos with mixed stacks.
Senior embedded-engineer code review. Defaults to Classic AUTOSAR embedded C and covers two concerns: (1) Correctness review — integer overflow, volatile correctness, ISR/task race conditions, stack usage, dynamic memory, control flow, AUTOSAR/ISO 26262 readiness, with findings rated Critical/Major/Minor; (2) Naming review — AUTOSAR Classic conventions for modules, SWC types, port names, runnables, C identifiers, typedefs, macros, and enumerations, aligned with MISRA Rule 5.x identifier uniqueness. Audits existing code and generates correctly-named identifiers for new work. Handles Adaptive AUTOSAR C++14+ when the input is C++ or names ara::, reviewing RAII, exceptions vs ara::core::Result, smart pointers, and AUTOSAR C++14 / MISRA C++:2023 naming. A third focus, legacy modernization assessment, proposes the smallest safe steps toward a MISRA-conformant, MCAL-abstracted, unit-tested state. Returns decision-ready findings with a built-in self-check and can emit a self-contained HTML report under analysis/.
Extract domain, architecture, business rules, workflows and a business glossary from an existing (often poorly documented) codebase, then validate the findings with a senior BA/Product Owner one question at a time. Produces onboarding-grade docs under docs/ that give a new team member — human or AI — enough context to be productive, ready for harness engineering / Spec Kit. Use when onboarding onto an unfamiliar codebase, reverse-engineering business knowledge, reconstructing lost documentation, or preparing a repo for spec-driven development.
When the user wants to manage regulatory compliance, ensure supply chain adherence to standards, or navigate trade regulations. Also use when the user mentions "regulatory compliance," "trade compliance," "customs," "import/export regulations," "certification management," "audit management," "REACH," "RoHS," "conflict minerals," "country of origin," or "trade restrictions." For sustainability compliance, see sustainable-sourcing. For quality standards, see quality-management.
Audit and manage the full project context landscape: CLAUDE.md memory hierarchy, project documentation, markdown footprint, and content overlap. Detects project type, scores quality, flags stale docs, and reports total context cost. Trigger with 'audit context', 'audit memory', 'update CLAUDE.md', 'restructure memory', 'session capture', 'check project docs', 'markdown footprint', or 'what docs does this project need'.
When the user wants to manage supplier contracts, negotiate terms, track compliance, or optimize contract performance. Also use when the user mentions "contract lifecycle," "contract negotiation," "SLA management," "contract compliance," "renewal management," "terms and conditions," "pricing clauses," or "contract analytics." For supplier selection, see supplier-selection. For spend analysis, see spend-analysis.
Use when the user needs to write, rewrite, review, or critique commercial copy meant to drive a reader toward action — landing pages, hero sections, headlines, email (cold outreach, nurture, promo, launch), CTAs, microcopy, product descriptions, value propositions, ads, or taglines. Also for copy audits, clarity checks, conversion reviews, or choosing a framework (AIDA, PAS, BAB, FAB, 4U's). Applies even when the user just says "write me an email" or "review this page." Not for channel/budget strategy (use channel-expert) or pricing structure (use pricing).
Master of Chinese copy enhancement, skilled in cross-border promotion, Xiaohongshu, WeChat official accounts, WeChat Moments, short videos, e-commerce product detail pages, ad placements, landing pages, sales pages, email marketing, brand copy, and boss-approved output. Use when the user asks to write copy, revise copy, polish, create titles, write introductions, write brand copy, write ads, write cross-border marketing copy, write Xiaohongshu posts, write WeChat official account posts, write WeChat Moments posts, write short video voiceovers, write product detail pages, improve click-through rate, improve conversion rate, run A/B tests, or for any text that needs to persuade, seed interest, close a sale, or drive action.
When the user corrects a factual error, immediately investigate its root cause. Don't just note the correction — trace the error to its source, fix the source, and prevent recurrence. Every factual error is either a data error (bad brain page, bad memory file, bad rendered SOUL/USER identity, bad facts row) or a hallucination (LLM confabulated from partial signals).
Author continuously-running online evaluations in PostHog AI observability, grounded in real failure modes you've identified. Use when the user wants evaluations that automatically score new generations or whole traces going forward — "create an eval to catch X", "continuously check that responses do Y", "turn these failures into evals", or after reading traces finds a recurring failure. Covers letting the explored data decide how many evals to create, proposing that set for the user to pick, choosing the target and eval type (hog / llm_judge / sentiment), configuring an llm_judge's provider and model, scoping which generations trigger it via conditions, creating disabled, verifying scope, enabling, and backfilling past data. Proposes a sentiment eval when no failure mode is worth catching. Finding and ranking the failure modes worth evaluating is its own job — use exploring-ai-failures first. To debug or manage evaluations that already exist, use exploring-llm-evaluations.
Reviews and critiques a CV/resume against proven best practices, providing a structured checklist with severity-tiered feedback and actionable suggestions for each item. Use this skill whenever a user asks to review, critique, check, improve, or get feedback on their CV or resume — even if they just say "look at my CV", "is my resume good?", "what's wrong with my resume?", or paste their CV content directly into the chat. Adapts feedback based on the user's experience level (student, mid-level, senior). Always use this skill when a CV or resume is present in the conversation.
Serverless compute for Databricks jobs, Lakeflow pipelines and Declarative Automation Bundles (DABs), with STANDARD or PERFORMANCE_OPTIMIZED and classic only where serverless cannot run the workload. Use when creating, deploying, scheduling or editing a job or pipeline, or deciding its compute. Invoke BEFORE writing a job spec. For migrating existing classic workloads, use databricks-serverless-migration.
Use when querying or calling pay-per-token foundation models, or when creating, updating, listing, or querying Unity Gateway (also called Unity AI Gateway) services, including model provider services, MCP services, system.ai services, and three-part Unity Catalog service names (catalog.schema.service). Also use for managing spend budgets and usage alerts on Unity Gateway, and for migrating from legacy AI Gateway. Also use when a workload with Model Serving endpoint permissions (for example CAN_QUERY) gets PERMISSION_DENIED querying a three-part model name, since serving-endpoint access does not grant access to a Unity Gateway service. Not for configuring legacy AI Gateway on Model Serving endpoints; use databricks-model-serving for that.
Audit temporal and lifecycle modeling — soft-delete consistency and its uniqueness/query traps, presence and integrity of an audit/history trail, and retention / GDPR right-to-erasure handling (hard delete vs anonymization, retention windows). Module M8. Feeds the Design & Integrity score (Temporal category).
Investigates application problems and incidents using Grafana metrics, logs, and traces via gcx: errors, latency, regressions, blast radius, and trace comparison with baselines. Accepts alert payloads, dashboard links, and trace IDs. For alert rule semantics, use investigate-alert.
Makes an agent crypto-native for onchain capital markets. Use for ANY question about DeFi, vaults, curators, yield, stablecoins, synthetic dollars, RWAs (real world assets), tokenized stocks, lending, perps, options, LP positions, token launchpads, bonding curves, or onchain credit. Trigger for learning ("what is a covered call", "explain post-only"), due diligence ("assess this vault", "is this APY sustainable"), trade anatomy ("what is this fund actually short"), curator comparisons, fee rails ("where does my gas fee go"), RWA mint and redeem mechanics ("is this tokenized APY real"), squeezes and manipulation ("is this a short squeeze", "is this wash traded"), crowding ("conviction or a crowded exit"), DeFi content tasks, and monitoring ("what changed this week"). Trigger even without the word DeFi when the subject is onchain yield, crypto tokens, rates, or market structure. Refresh live data before anything numeric. Not for TradFi-only rates or credit questions, LLM tokens, or transaction execution.
Parallelizes work across six patterns — built-in agents, worktree-isolated agents, background agents, custom inline-brief agents, multi-turn agent teams, and parallel task dispatch — choosing teams when agents must share state, subagents when independent. USE WHEN parallel execution, agent team, swarm, spawn agents, fan out, divide and conquer, multi-agent, coordinate agents, custom agents.
Find the mocks, stubs, fakes, hidden fallbacks and placeholder implementations standing on a real code path outside the tests, where nothing real runs and a success report is false, and propose the real dependency or an honest failure for each. Use when a change or a named file, directory or diff may ship a test double or a swallowed error into production, when a catch or except returns an empty list, None, a default or true, when a guard lets a request through because its key, config or input is missing, when a function raises NotImplementedError or returns a TODO value on a live path, when a feature reports success but nothing seems to happen, or before calling work done that touched error handling, integrations or service clients.
Translate a design (Figma mockup, spec, screenshot, or verbal description) into code with pixel- and motion-level fidelity, using the browser's rendering pipeline as a constraint: exact spacing/timing via tokens, GPU-safe animations, no layout thrashing, 60fps interactions. Use when the user says 'implement this design', 'make it match the mockup', 'it doesn't look like the Figma', 'the animation stutters/drops frames', 'set up design tokens', or when a diff between design and implementation needs to be closed. Produces the faithful implementation plus a fidelity report (design value → implemented value → any deviation and why) and performance notes.
Generate a DESIGN.md design-system spec (color tokens, typography, spacing, shadows, components) and optional HTML preview from any website URL using the designmdme CLI. Use when the user wants to extract, capture, replicate, or reference a website's design system or visual style — e.g. "build a UI like stripe.com", "match this brand's look", "get the design tokens for linear.app", "make my app look like notion.so", or any time you need an accurate color palette, font stack, or component styling from a real site before writing UI code.
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.