Software Engineer

Write production-ready code with clean architecture, proper error handling, and pragmatic trade-offs between shipping fast and building right.

This skill's content matches its purpose and asks for nothing sensitive, but take the following precautions before installing or invoking it autonomously: - Inspect the SKILL.md and included files in a text editor that can show invisible/control characters; remove any unexpected unicode control characters. - If you plan to allow autonomous execution, run the skill first in a restricted/sandboxed agent with no network access and limited file permissions to verify behavior. - Because the files assert they won't make network requests or store data, don't treat that as enforcement—ensure the runtime environment enforces the same restrictions if you need them. - If you don't need autonomous invocation, disable it (or limit the skill's privileges) until you've validated the content. If you want, I can display the SKILL.md with non-printing characters highlighted or produce a cleaned version with control characters removed.

Type: OpenClaw Skill Name: software-engineer Version: 1.0.0 The OpenClaw AgentSkills skill bundle for 'software-engineer' is benign. All files (`_meta.json`, `SKILL.md`, `architecture.md`, `patterns.md`, `testing.md`) contain only metadata and documentation outlining software engineering best practices. The `SKILL.md` explicitly states that the skill does NOT 'Store any data externally', 'Make network requests', or 'Access files outside the current project', and there are no hidden instructions or prompt injection attempts to contradict these claims. The content is purely instructional for an AI agent and lacks any malicious code or risky capabilities.