Skip to main content
EdgeSpark is designed for AI agents as a first-class user, and EdgeSpark supports all AI agents. This section explains the platform from the agent’s perspective: what the platform enforces on your behalf, how to work autonomously, and how the first-class plugin paths differ from the generic skills + MCP path that supports every other agent.

​
What’s in this section

The harness model

What EdgeSpark enforces and why. The guardrails that prevent common mistakes.

Declarative workflow

How to pull schema, write code against types, and deploy iteratively.

Deploy and test loop

How to iterate safely without a local dev server while staging is still coming soon.

Handling errors

How to read deployment errors, type errors, and runtime failures.

Minimal human input

Keep going by default and stop only for login, secret entry, or destructive actions.

AGENTS.md reference

How to read and follow the project’s agent instructions file.

Supported agents

Claude Code, Gemini CLI, OpenAI Codex, GitHub Copilot, Cursor, OpenCode, Amp, Devin, Aider, Windsurf, Cline, Continue, Antigravity, Kiro, and all AI agents.

Platform limits

Request size, database row limits, storage caps, and all platform constraints.
Last modified on April 9, 2026