The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by reading git commit logs, diffs, and existing Jira issue descriptions. This data is interpolated into the agent's context to generate summaries and test cases. While the skill includes a manual confirmation step (Step 3) for the user to review the draft, it lacks specific instructions to sanitize these inputs or use boundary markers to prevent the agent from obeying instructions hidden within commit messages or issue descriptions.
Ingestion points: git log, git diff, and git show output in Step 1 and 2 of SKILL.md; existing Jira issue descriptions fetched in Step 4.
Boundary markers: Absent. The instructions do not specify the use of delimiters or instructions to ignore embedded commands when processing the external text.
Capability inventory: The skill has capabilities to execute shell commands (git) and perform authenticated network writes to Jira via the Atlassian MCP (update-issue, create-issue).
Sanitization: Absent. No filtering or escaping of the ingested data is defined.
[COMMAND_EXECUTION]: The skill utilizes several local shell commands to analyze the repository state, including git fetch, git log, git diff, and git show. These commands are standard for the skill's stated purpose of reviewing code changes and are constrained to the local git environment.

1k-jira-qa-handoff