Generate candidate fixes for verified security findings. Consumes TRIAGE.json (preferred), VULN-FINDINGS.json, INCIDENTS.json, or a vuln-pipeline results directory. Pipeline input is delegated to the execution-verified `vuln-pipeline patch` ladder; static-analysis input gets a per-finding patch subagent + independent reviewer and is written as inert diffs for human review.
安装
npx skills add anthropics/defending-code-reference-harness --skill patchnpx skills add anthropics/defending-code-reference-harness --skill patch -a claude-code -gnpx skills add anthropics/defending-code-reference-harness --skill patch -a codex -ggemini skills install https://github.com/anthropics/defending-code-reference-harness.git --path .claude/skills/patchgh skill install anthropics/defending-code-reference-harness patchgit clone --depth 1 https://github.com/anthropics/defending-code-reference-harness /tmp/anthropics__defending-code-reference-harness
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/anthropics__defending-code-reference-harness/.claude/skills/patch ~/.claude/skills/patch # Claude Code
cp -R /tmp/anthropics__defending-code-reference-harness/.claude/skills/patch ~/.agents/skills/patch # Codex, Cursor, Gemini CLI, Copilot, OpenCode