The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using user-supplied input ($ARGUMENTS) in git log and git diff operations. While this is the primary functionality of the skill, it presents a risk of command injection if the execution environment does not properly sanitize these arguments.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the repository, including commit messages, code changes, and file contents, to generate narratives. This data could contain malicious instructions designed to manipulate the agent's output.
Ingestion points: Git logs (git log), diffs (git diff, git show), and file content reading (Step 2c).
Boundary markers: The instructions do not define clear delimiters or specific warnings to ignore instructions embedded within the code or git history being analyzed.
Capability inventory: The agent has the ability to execute shell commands and write session records to the local .learn/ directory.
Sanitization: No escaping or filtering of the ingested content is specified before it is processed for narrative generation.

diff-story