The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/herdr-tab-peer.sh utilizes dynamic execution patterns to process JSON data and resolve workspaces. Specifically, the jqget function employs Python's eval() on arguments passed from the shell. Additionally, the script embeds and executes multi-line Python snippets via python3 -c at runtime to perform filtering and matching logic.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external instructions via the --prompt-file argument or standard input, which are then used as bootstrap prompts for newly created peer agents. This creates a surface where unvetted data can influence agent behavior.
Ingestion points: The PROMPT_FILE parameter in scripts/herdr-tab-peer.sh reads content from user-specified file paths or stdin.
Boundary markers: There are no explicit markers or guardrails implemented to delimit or sanitize the ingested prompt content.
Capability inventory: The skill has the capability to execute herdr CLI commands to create system tabs, start new agents, and resolve workspace configurations.
Sanitization: The script does not perform sanitization or validation on the content of the prompt file before passing it to the herdr agent prompt command.
[COMMAND_EXECUTION]: The skill relies on executing local shell scripts (check-env.sh, herdr-tab-peer.sh) and making extensive use of the herdr CLI tool to manage terminal environment state, including support for remote command execution on SSH-connected machines via the --machine flag.

to-peer