For AI Agents

Push Chain is designed for AI agents, copilots, and automation systems to both understand and execute actions across chains, with minimal ambiguity and maximum reliability.

The resources below form a layered integration stack: discovery → capabilities → execution.

Entry Points

Resource	Purpose
/llms.txt	Compact structured discovery file for LLMs. Start here to understand Push Chain and locate the right resources.
/agents/	Machine-readable interface layer: capabilities, workflows, schemas, and code examples for execution.
/llms-full.txt	Full documentation corpus in plain text. Ideal for deep context, retrieval systems, and long-context reasoning.

What This Enables

• Deterministic execution: structured workflows eliminate guesswork when planning transactions
• Reduced ambiguity: typed schemas and capability definitions map directly to SDK calls
• Intent → capability → execution: a clear path from natural language to on-chain action
• Reliable AI integration: works with any agent framework, RAG pipeline, or developer copilot

Use these endpoints to let AI agents understand and execute Push Chain transactions natively — no unstructured scraping required.

Agent Interface Structure

The /agents/ layer is organized into machine-readable files that build on each other:

File	What it contains
/agents/index.json	Discovery map — lists every file, its purpose, and where to start
/agents/capabilities.json	Every SDK capability with inputs, outputs, and signatures
/agents/workflows/index.json	Step-by-step execution guides for common tasks
/agents/schemas/index.json	JSON schemas for all request/response types
/agents/examples/index.json	60+ minimal, executable TypeScript code snippets
/agents/decision-tree.json	Branching logic to select the right capability from intent
/agents/retrieval-map.json	Maps tasks to authoritative documentation sources (for RAG)
/agents/errors.json	Error catalog with recovery actions

Integration Paths

Human developer path

1. Follow the Quickstart to run your first transaction
2. Work through Tutorials for end-to-end flows
3. Use the SDK and UI Kit for production integration

AI agent path

1. Fetch /llms.txt as the entry layer
2. Load /agents/index.json to discover all capabilities and workflows
3. Use /agents/decision-tree.json to map user intent to the right capability
4. Execute using the matching workflow from /agents/workflows/index.json
5. Validate with schemas from /agents/schemas/index.json
6. Fetch /llms-full.txt when full context is needed

Notes

• Prefer canonical workflows in /agents/workflows/ over navigating raw documentation
• Use /agents/schemas/ for precise input validation before execution
• Treat /llms.txt as the entry layer and /agents/ as the execution layer
• /agents/retrieval-map.json maps every capability to its normative documentation source — use it to ground responses
• All code examples in /agents/examples/ are minimal, self-contained, and ready to execute