The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/validate_qt_patch.py executes arbitrary shell commands provided via the --cmd argument using subprocess.run(shell=True) for build and test validation.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text from external Jira issues and Gerrit comments to generate commit messages and suggested replies.
Ingestion points: Jira issue fields in scripts/scan_qtbug.py and review comments in scripts/review_analyzer.py.
Boundary markers: None present; external data is interpolated directly into prompts and messages.
Capability inventory: Subprocess shell execution, Git operations, file system writes, and network operations.
Sanitization: No sanitization or filtering of external content is implemented.- [PERSISTENCE]: The scripts/qtbug_worktree_manager.py script appends host keys for codereview.qt-project.org to the user's ~/.ssh/known_hosts file using ssh-keyscan.- [EXTERNAL_DOWNLOADS]: Fetches and installs the official Gerrit commit-msg hook from codereview.qt-project.org using curl and urllib.request. (Source is recognized as a well-known service).- [DYNAMIC_EXECUTION]: Uses subprocess.run across multiple scripts to dynamically orchestrate Git commands and workflow scripts.

qtbug-upstream-fixer