The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs shell commands by interpolating variables derived from external data without sufficient sanitization.
The variable $org is interpolated directly into the command releases admin org get "$org" within a bash loop in Step 2 of the documentation.
The variable {slug} is interpolated into sub-agent prompt templates and fallback commands in Steps 4 and 5 (e.g., releases admin playbook {slug} --notes-file -). If these values (slugs or org names) are derived from untrusted sources and contain shell metacharacters, it could result in arbitrary code execution in the local environment.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection due to the way it handles external data.
Ingestion points: The skill retrieves and processes organization metadata, source details, and release logs through the releases CLI in both the Compilation and Verified workflows.
Boundary markers: The prompt templates for sub-agents (Step 4) do not include clear boundary markers or instructions to ignore potential commands embedded within the processed data.
Capability inventory: The agents and sub-agents have the ability to execute shell commands and modify production data (playbooks) using the releases CLI.
Sanitization: There is no evidence of data sanitization, validation, or escaping before the retrieved content is interpolated into agent prompts or executed in shell scripts.

seeding-playbooks