The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill employs curl | python3 pipelines to process data from external sources like the Wayback Machine. Critically, it provides specific instructions on how to circumvent security scanner detections of these patterns by using temporary files.
[COMMAND_EXECUTION]: The skill makes extensive use of the terminal tool to execute curl, proxychains4, and python3 scripts. It includes strategies to bypass heredoc limitations and automated security blocks for shell commands.
[DATA_EXFILTRATION]: The skill's primary function is deep reconnaissance, specifically targeting sensitive information such as .env files, .git/config, wp-config.php.bak, and searching for API keys and JWTs across multiple target environments.
[PROMPT_INJECTION]: The 'Skill Evolution' task implements an indirect prompt injection surface by taking data collected from untrusted external targets and using it to dynamically generate or patch the agent's skills via the skill_manage tool.
[PERSISTENCE]: The skill uses the cronjob tool to establish an 'eternal' orchestration loop, ensuring the reconnaissance activities persist autonomously over time.
[OBFUSCATION]: The instructions suggest using decimal IP encoding (e.g., converting 169.254.169.254 to 2852039166) as a technique to bypass security filters that block raw IP literals in execution contexts.

parallel-recon-triad