The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages Proxyman via shell commands executed through the agent host. To mitigate risks, it strictly instructs the agent to discover syntax dynamically from the local installation's help menus rather than relying on static or guessed commands.- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill facilitates the export of logs and configuration files to the local filesystem. It addresses security risks by requiring user confirmation before overwriting files and providing clear warnings that logs may contain unredacted secrets.- [CREDENTIALS_UNSAFE]: While interacting with license activation commands, the skill enforces a 'no-secrets' policy where the agent provides command templates with placeholders like <LICENSE_KEY>. It explicitly forbids the execution of commands containing real secrets to prevent them from being logged in shell history or process listings.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes external data such as captured network logs and configuration files during import/export operations.
Ingestion points: Reading configuration files during import and processing log data for export-log (SKILL.md).
Boundary markers: The instructions mandate redacting secrets and require explicit human confirmation for all destructive or high-impact actions.
Capability inventory: Shell command execution for proxyman-cli and filesystem write access via export commands.
Sanitization: Employs mandatory output sanitization and verification of system state after command execution.

proxyman-cli