Claude Security
Deep vulnerability scan of your own code, with every finding challenged before it is reported.
Install
OfficialClaude Code: /plugin marketplace add anthropics/claude-plugins-official → /plugin install claude-security@claude-plugins-official
Yayına çıkmadan önce kendi kod tabanında güvenlik açığı taraması yapılırken
- Author
- Anthropic
The failure mode of an AI security review is confident noise: plausible findings that do not reproduce, which train the team to ignore the report. This plugin puts a verification pass between discovery and output — each candidate finding is attacked before it survives, and the tally is computed in code rather than asserted by the model. Effort is selectable, so a quick pass and a deep audit are the same tool at different settings, run entirely inside the session.
Similar skills
Code Review
SkillMulti-agent pull request review with confidence scoring to filter out false positives.
Security Guidance
SkillSecurity review of agent-written code at three points: on edit, on stop, and at commit.
CodeRabbit Code Review
SkillRuns CodeRabbit's reviewer from inside the agent and applies its PR feedback change by change.
Comments(0)
Sign in to comment