The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions specify checking for and utilizing the .jira-token file to authenticate against Jira. Accessing sensitive credential files, even for primary functionality, represents a data exposure risk.
[COMMAND_EXECUTION]: The skill utilizes several shell commands to manage the local development environment, including lsof -ti:<PORT> | xargs kill -9 to clear network ports and ulimit -n 65536 to modify system file descriptor limits. It also executes npx playwright test on dynamically generated test scripts to verify bug reproductions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it fetches untrusted data from Jira and GitHub issues (summaries, descriptions, and reproduction steps) and uses this content to generate executable Playwright tests and perform root-cause diagnosis without explicit sanitization.
Ingestion points: Fetches issue titles, bodies, and comments from the GitHub CLI (gh issue view) and Jira API/MCP in Step 1.
Boundary markers: The skill lacks explicit delimiters or safety instructions when interpolating external issue content into the agent's prompts or test templates.
Capability inventory: The skill has the ability to write files to the local filesystem, execute browser tests via Playwright, and create pull requests on GitHub.
Sanitization: No validation or filtering of the fetched issue content is described before the data is used to drive agent decisions or code generation.

bug-fix