The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform repository metadata checks using the GitHub CLI (gh repo view) and to manage local state by reading configuration files (cat .adr-session.json). These operations are consistent with the skill's intended purpose of technical evaluation.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied technology names and repository identifiers, which are then interpolated into shell commands for verification purposes. This creates a potential attack surface for command injection if malicious strings are provided.
Ingestion points: User-provided technology options and repository paths defined during the framing step in SKILL.md and utilized in references/decision-archetypes.md and references/decision-preferred-patterns.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompts used for command generation.
Capability inventory: The skill has access to the Bash tool for repository inspection and local file operations, as well as the Write tool for updating project plans.
Sanitization: The instructions do not specify any sanitization, escaping, or validation logic for the external strings before they are interpolated into shell command templates.

decision-helper