The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using git log in SKILL.md (Step 3) and references/daily-detail.md. These commands interpolate variables <N> (card number) and <slug> (branch/slug name) which are derived from external data sources like GitHub, Linear, or Jira. If an attacker controls the card title or branch name, they could inject shell commands into the execution context.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from external sources to generate summaries and recommendations.
Ingestion points: Card data (title, status, comments) returned by the read_board action in SKILL.md Step 1.
Boundary markers: The instructions do not specify any delimiters or safety prompts to isolate external data from the core logic.
Capability inventory: Shell execution via git log and the ability to trigger other mutating skills like board-superpowers:intaking-requirement.
Sanitization: There is no mention of input validation or character escaping for data retrieved from the project management tools.

briefing-daily