The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which serves as a potential attack surface for prompt injection.
Ingestion points: The skill instructions in SKILL.md (Step 2) direct the agent to "Read the ticket and relevant task artifacts" and "Read the complete pull request diff."
Boundary markers: There are no instructions to use delimiters or ignore embedded instructions within the ingested data.
Capability inventory: The skill has the ability to execute shell commands, run local scripts, and write content to GitHub pull requests via the gh CLI.
Sanitization: The skill does not implement sanitization or filtering for the data read from tickets or diffs before incorporating it into the final output or decision-making process.
[COMMAND_EXECUTION]: The skill executes local shell commands and a provided bash script (scripts/publish-comment.sh) to interact with the file system and GitHub API.
Evidence: SKILL.md contains multiple bash blocks using gh, git, jq, and mktemp. It also invokes $SKILL_PATH/scripts/publish-comment.sh with arguments derived from the environment and agent context.

visual-pr