Evolve SDK provides language-specific references, runtime patterns, and tooling to build, run and observe CLI-style agent applications in secure cloud sandboxes. It codifies recommended workflows for TypeScript and Python, including run/kill lifecycle, streaming events, structured outputs, checkpointing, and storage clients. The skill collects best-practice docs and examples that help developers scaffold agent apps, manage sessions, and integrate with MCP servers and composable tool routers.
Use this skill when you are creating or debugging agent-driven CLI apps, configuring sandbox providers, building Swarm or Pipeline patterns (map/filter/reduce/bestOf), implementing structured output schemas, or instrumenting session-level observability and cost tracking. It's also useful when you need language-specific templates (TS or PY) for agent configuration and safe cleanup.
Best fit for agent stacks that support CLI-style runtimes and structured outputs (e.g., Claude Code / Codex / Gemini CLI / other code-oriented agents).
Evolve SDK skill provides comprehensive dual-language (TypeScript & Python) guides for building CLI agent applications in secure cloud sandboxes. No bundled scripts to test — the skill is purely reference-based with progressive disclosure of SDK documentation. The SKILL.md is well-structured with clear triggers, language detection logic, and a detailed topic index covering installation through advanced Swarm/Pipeline patterns.
Well-crafted SDK documentation skill. The self-update mechanism using git sparse-checkout is reasonable but runs without confirmation (-8). No scripts present, so security assessment is based purely on SKILL.md content. The dual-language topic index is exemplary — best-in-class progressive disclosure. Usefulness is moderate because the audience is limited to Evolve SDK users specifically, though the SDK itself covers a growing space (sandbox agents).