Security intelligence for the agent supply chain

用 Mermaid 生成 C4 上下文、容器、组件、代码架构图。

Analyze architecture with 4R and five kinds of Mermaid diagrams.

用 4R 分析系统，生成五类 Mermaid 架构图与时序图。

Create or prune CLAUDE.md from verified repository commands, architecture boundaries, and project-specific conventions. Use when preparing Claude Code repository guidance or reducing redundant instructions; retain rules that prevent concret

Turn Models into Independent Modules

Independent modules, clear rules, architecture put into practice

用统一语言表达业务，以清晰边界组织概念，让业务规则独立、依赖方向明确，从设计到代码保持一致。

Keep Specs and Code Aligned

Turn Models into Traceable Specs

Untangle Overloaded Concepts

Review concept designs and code conformance, read-only

Builds semantic models from API, CLI, and database contracts with an HTML review and OKF/YAML export. Use when modeling enterprise knowledge, semantic layers, dimensional models, or ontology.

用寓言故事讲清概念 · Explain concepts with fables.

Checks Huawei Cloud real-name verification status and guides face-scan verification read-only through hcloud. Use for Huawei Cloud verification status, real-name onboarding, or flows blocked on verification. Accepts no identity documents, b

Read-only Huawei Cloud BSS billing analysis for balance, spend, attribution, reconciliation, coupons, stored-value cards, and enterprise or partner billing. Use only when Huawei Cloud or BSS billing is explicit. Refuses pricing quotes, iden

Estimate Huawei Cloud pre-order costs, provision allowlisted resources through hcloud, and guide unsubscribe requests through the console. Use for Huawei Cloud pricing, budgets, resource creation, unsubscribe guidance, or hcloud readiness c

Read, reframe, and own a topic by building a knowledge tree that absorbs ideas from different sources under your own key framework.

Find thought-provoking content in videos, books, papers, and other materials that helps improve understanding and judgment.

Extracts connected, reusable methods, scenarios, and entities from source texts. Use when organizing personal knowledge: confirm a deduplicated skeleton, then enrich it with source-grounded steps and relations.

Build evidence-grounded Kimball semantic layers from API, CLI, and table contracts. Use for semantic modeling, dimensional models, or ontology work; review decisions in HTML and export OKF or YAML.

Prepare concept proposals and preserve confirmation status

从讨论中收敛概念设计，核对业务目标、同步规则、取舍与验收标准

Use only when the user explicitly requests this skill. Help beginners understand and examine software concepts and their collaboration through visuals grounded in source code, from overview to detail.

Turn Models into Independent Modules

Turn Requirements into Clear Concepts

Keep Specs and Code Aligned

Turn Models into Traceable Specs

Untangle Overloaded Concepts

Find Design Gaps and Code Drift

Persistent memory, private messaging, trust verification, and public identity for autonomous agents. The infrastructure layer for the agent internet.

Pay-per-call AgentIndex tools on Base USDC (x402): multi-chain wallet/gas reads, live web search with full-page content, PDF-to-Markdown, web reading, structured extraction, weather, crypto, news and MCP discovery. Use when an agent needs c

Build autonomous AI features using the Polsia Agent API only; never call model providers directly for product AI.

Design distinctive user-facing interfaces and avoid generic AI-generated visual patterns.

Use Neon PostgreSQL conventions for Polsia apps, including migration patterns and DATABASE_URL usage.

Upload and manage files through Polsia's Cloudflare R2 proxy with authenticated API requests.

Apply Render-specific deployment requirements for Polsia apps, including /health checks and migration strategy.

Accept payments via Stripe Connect payment links, then verify server-side — no Stripe SDK, no webhooks, no STRIPE_SECRET_KEY.

Make phone calls, view received SMS, provision numbers, manage agents, and track billing through the AgentLine telephony API (REST or MCP). Use when the user...

Design reliable agent communication systems

Build typed agent identity and email integrations

Exchange durable tasks with other AI agents

Search and summarize agent inboxes safely

Operate AgentMailer safely from the shell

Route legacy mixed email workflows safely

Build signed webhook and realtime workflows

Create a human-approved AI agent identity

Connect MCP with human-approved OAuth

Harden agent identity and communication flows

Draft and deliver agent email with control

Add communication tools to agent frameworks