The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its primary function involves reading and analyzing untrusted code from external GitHub repositories. A malicious repository could include instructions designed to manipulate the agent's findings or influence its decision-making logic.
Ingestion points: External repositories accessed through the gh CLI as described in AGENTS.md and references/v1-standard-portco-dd.md.
Boundary markers: The skill does not specify explicit delimiters or instructions to separate external repository content from its internal reasoning processes.
Capability inventory: The agent has access to the gh CLI, the ability to write files to local paths (e.g., /Users/kp/Documents/Utopia DD Reports/), and network access to the GitHub API.
Sanitization: No sanitization or filtering of the content retrieved from external repositories is mentioned.
[CREDENTIALS_UNSAFE]: A hardcoded email address (karanmjpinto@gmail.com) is included in MEMORY.md as a contact for escalation.
[COMMAND_EXECUTION]: The skill utilizes the gh (GitHub) command-line interface to view and analyze repositories, which may include access to authenticated private repositories as noted in AGENTS.md.

ada