.Agent

Tool design, binding, and error handling patterns for LangChain agents

Web scraping with agents, browser automation, content extraction, and ethical scraping practices

CDS view creation (basic, composite, consumption), associations, compositions, CDS annotations (semantic, UI, analytics), CDS access control (DCL)

Continuous CI/CD pipeline monitoring with automatic error detection and fix loop

Guide users through selecting appropriate MCP servers during project generation

Static code analysis with AST parsing, complexity metrics, dependency graphs, and quality scoring

Configure project management system with backend and methodology selection

Behavior-driven development with Gherkin feature files, scenario outlines, and step definitions

Automated governance, hook installation, pre-commit validation, branch isolation, and safe commit operations.

LangGraph state schemas, reducers, checkpointing, and persistence

Create implementation task from story or description

Image-to-text with various engines (Tesseract, EasyOCR, cloud APIs), PDF text extraction, table recognition, and layout analysis

Production model serving with vLLM, TGI, Ollama, Triton, FastAPI endpoints, batching, and GPU optimization

Unit testing with mocks, integration testing, LangSmith evaluation, benchmarking

Use when you have a spec or requirements for a multi-step task, before touching code. Enforces bite-sized task granularity.

Retry strategies with tenacity, fallback patterns, circuit breakers, graceful degradation

Model Context Protocol overview, server creation, tool exposure, and common integrations

Manage the Antigravity Dual-Storage Memory system (SQL + Vector) and implement cognitive lifecycle hooks.

Create user story from description with acceptance criteria

Verify internal and external links in project documentation

Guide users through articulating why their agent system should exist, who it serves, and how success will be measured

BTP Cloud Foundry deployment (cf push, MTA), service bindings (XSUAA, destination, connectivity), Kyma runtime deployment, SAP Build Work Zone integration

Interactive guidance for selecting appropriate trust verification tier

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Standardized agent creation and evaluation with mandatory schema validation

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

Platform Application Bundle Protocol (PABP) management, MCP config generators, bundle importers, and compliance verifiers.

Send emails using the cached Google Workspace credentials.

Use when implementing any feature or bugfix, before writing implementation code. Enforces the Iron Law of TDD.

On-demand version audits using version-registry.json as source of truth

Agent definition file generation with mandatory schema validation against schemas/agent.schema.json

Speech-to-text (Whisper, cloud APIs), text-to-speech synthesis, real-time transcription, and audio analysis

Receive and apply updates from the Antigravity Agent Factory including skills, knowledge, and agents

Research existing solutions before building to create multiplied value

Convert tests between frameworks (pytest, unittest, behave) and translate test patterns across languages

REST API design (resource naming, HTTP methods, status codes), OpenAPI/Swagger documentation, API versioning strategies (URL, header, query), GraphQL schema design, gRPC proto definitions, Pagination patterns, Error response format, Rate limiting headers

Technology stack selection and configuration skill

Fast-track 10-15 minute onboarding for individual developers with guided prompts and celebrations

LLM response caching, semantic caching, memoization, cache invalidation

Systematic detection and fixing of CI/CD pipeline failures.

Interactive skill for reviewing, applying, and managing knowledge updates

Validates that project structure documented in README.md accurately reflects the actual filesystem structure

Standardized agent pattern creation and validation

Tactical Blueprint for AI Security and Guardrails. Focuses on prompt injection defense, PII detection, and content safety for agentic systems.

PyTorch training loops, distributed training with DeepSpeed/FSDP, hyperparameter tuning with Optuna, experiment tracking, mixed precision training, checkpointing

LangChain 1.x patterns for chains, tools, memory, and structured outputs

Produces a complete Agentic PRD from the approved prototype brief. Combines human-readable product context with machine-executable acceptance criteria, structured story format, and explicit AI-component requirements. Use when the user wants to write requirements, create a PRD, document what to build, or says "let's write the spec", "document the requirements", or "what are we building". Requires knowledge/prototype-brief.md to be present and approved. Writes to knowledge/prd.md.

GitHub operations and repository management using the GitHub MCP Server

Safe commit and release workflow with auto-sync, changelog updates, and learning from failures

Interactive configuration and onboarding skill for Antigravity Agent Factory settings

Export learning workshops to standalone projects with all necessary artifacts for maximum AI-assisted learning

Extend Factory knowledge, skills, and templates via research, documents, or user links

Create and extend workflow definitions using the workflow system architecture

Enforcement of safety guardrails, axiom verification, secret scanning, and mutability protections.

Precise skill for updating JSON registries, dependency graphs, and manifests.

Multi-index RAG, query decomposition, contextual compression, and self-querying retrievers

Input validation and sanitization, output filtering, code execution safety, and permission management

Guide users through selecting practices that maintain excellence, alignment, and continuous improvement

Structures an ambiguous idea or pain point into a crisp, bounded problem statement with user context, assumptions, success criteria, and out-of-scope boundaries. Use this skill at the very start of any ideation session, when the user describes a vague goal, a frustration, an opportunity, or says something like "I want to build X" or "users are struggling with Y". Invoke before brainstorming — framing must precede divergence.

Fetch and manage up-to-date API documentation for coding tasks using Context Hub (chub). Ensures agents have situation-adequate documentation for modern tools and prevents hallucinations.