Aikido Security
Scans the files an agent just wrote and loops on the fix until the findings actually clear.
Install
OfficialClaude Code: /plugin marketplace add anthropics/claude-plugins-official → /plugin install aikido@claude-plugins-official
Ajan kod yazdıktan ya da değiştirdikten hemen sonra o dosyaları açık ve sızmış sır için tararken
- Author
- Aikido Security
Three small skills cover setup, scanning, and reading the existing findings feed from the account. The scan skill is written for the moment right after code generation: it gathers every file touched in the session, sends contents in batches capped at fifty files per request, and reports vulnerable patterns and exposed secrets with file and line. What keeps it from spinning forever is an explicit stop rule — fix and rescan at most three times, and stop earlier if the remaining output can be argued as a false positive. Everything runs through the Aikido MCP server, so when the server is missing the skill says so with a setup pointer rather than quietly reporting a clean result.
Similar skills
SonarQube
SkillPuts SonarQube's rule set, secrets scanning and quality gate inside the agent's own edit loop.
Comprehensive Review
SkillA phased review that writes each stage to disk so later phases read files, not context memory.
Signed Audit Trails
SkillA single cookbook skill that explains signed tool-call receipts before you commit to running them.
Comments(0)
Sign in to comment