The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from git commit logs and GitHub Pull Request titles during regression analysis and release note generation.
Ingestion points: Data enters via git log and gh pr list commands in the 'Find the change' and 'Release notes mode' sections.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompt templates.
Capability inventory: The agent has capabilities to execute shell commands (git, pytest, gh, grep) and write files to the local directory.
Sanitization: There is no evidence of sanitization or filtering of the commit/PR content before it is processed or summarized.
[COMMAND_EXECUTION]: The skill utilizes local shell commands to automate development workflows.
Evidence: The skill instructs the agent to use git bisect run pytest to identify buggy commits, which executes arbitrary code contained within the project's test files.

regression-postmortem