Claude Code skills Ported to OpenCode

Moved the command and sub-agent definitions from anthropics/claude-code/plugins/, anthropics/skills, and anthropics/claude-code-security-review into the SKILL.md format for OpenCode.

The core result is 11 skills, covering code review, security auditing, feature development, frontend design, writing MCP servers, and AGENTS.md/CLAUDE.md management.

• Each skill is also provided as a slash command, such as /code-review and /security-review.
• To reduce the problem of local models outputting meta-instructions verbatim, the skill workflow was inlined directly into the slash command body.
• The loader approach of registering the skills directory via config.skills.paths.push() was based on the obra/superpowers pattern.
• The review skill incorporates additional verification perspectives, cross-checks for falsification, and mandatory reproduction scenarios to compensate for local models' tendency toward hasty judgments.

The repository is waybarrios/opencode-power-pack, and it can be seen as an example of moving the Claude Code-specific structure of commands/ and agents/ into the more general-purpose skills/.