The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute a local shell script scripts/build_review_packet.py with arguments (--intent and --include) populated directly from user-provided descriptions and file paths. If the agent does not properly escape these strings before execution, it allows for shell command injection.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes external, untrusted data including git diffs and requirement specifications, which serves as a vector for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via git diff outputs and external files passed to the packet-building script.
Boundary markers: The skill does not define delimiters or instructions for the agent or its subagents to ignore potential instructions embedded within the code changes or specifications.
Capability inventory: The skill possesses the capability to execute shell commands (git, python) and dispatch multiple parallel subagents (Agent tool).
Sanitization: No sanitization, validation, or filtering of the ingested content is performed before analysis.

code-review