The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Git commands to inspect the repository state. The instructions direct the agent to run git status, git diff --stat, and git log --oneline -15 to collect evidence for the handoff document.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository and conversation history to generate its output, presenting a surface for indirect injection.
Ingestion points: The skill reads external data via git status, git diff, and git log, and incorporates quotes from the user's transcript.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potentially malicious content embedded within the Git logs or transcript data.
Capability inventory: The skill has the capability to write files to the repository (e.g., HANDOFF.md or dated files in docs/handoff/).
Sanitization: No sanitization, escaping, or validation of the ingested Git output or transcript text is defined before it is written to the handoff file.

session-handoff