Automate authorized AI red-team reconnaissance across four priority classes.
npx skills add https://github.com/Repello-AI/artemis-recon --skill mentor-mode-repello-ai
Please help me install this Agent Skill. Skill: mentor-mode Source: https://github.com/Repello-AI/artemis-recon/tree/main/skills/mentor-mode Command: npx skills add https://github.com/Repello-AI/artemis-recon --skill mentor-mode-repello-ai
This Skill automates the process of authorized AI red-team reconnaissance, streamlining the discovery and assessment of AI assets in applications or products.
Start a new recon engagement for the chatbot 'chatbotA' using the /recon command.
High-intent search queries and answers about installing and using this skill.
Authorized AI red-team reconnaissance is the process of probing AI assets through legitimate multi-turn conversations to observe system prompts, inventory tools, evaluate competitor handling, and identify AI identity leakage.
You can automate AI red-team reconnaissance by initiating a multi-turn probing session using the /recon command, which steers the engagement across four priority classes to assess the target AI asset.
Yes, performing these automated AI security assessments requires both the Mentor controller and Claude Code, as they are necessary dependencies for session steering and executing the reconnaissance logic.
An AI asset analysis evaluates four priority classes: system-prompt observation, tool inventory, parroting and competitor handling, and AI identity leakage to discover potential security weaknesses.
This approach automates the reconnaissance phase rather than manual penetration testing, streamlining the discovery and assessment of AI assets through legitimate-conversation probing under the Mentor controller.
pdparchitectAudit Okta tenants for authentication, MFA, and API misconfigurations.
autorundevAudit AI applications for prompt injection, data exfiltration, and unauthorized tool usage.
shobhit-16Integrate conversation UX patterns for AI chat interfaces with branching threads.
Pa55w0rdEnumerate attack surfaces and guide security assessments for web and AI targets.
