The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/core/jira-validate.py includes a recommendation for users to install the uv package manager using curl -LsSf https://astral.sh/uv/install.sh | sh. While uv is a reputable tool from Astral, piped-to-shell execution is a risky installation pattern.
[COMMAND_EXECUTION]: In scripts/core/jira-validate.py, the skill uses subprocess.run to execute the command uv --version. This is a controlled execution used solely to verify that the required runtime dependency is correctly installed.
[PROMPT_INJECTION]: The skill processes unstructured external data from Jira issues, such as descriptions and comments, in scripts/core/jira-issue.py and scripts/workflow/jira-comment.py. This creates a surface for indirect prompt injection if a ticket contains malicious instructions.
Ingestion points: Data is ingested via scripts/core/jira-issue.py (issue fields) and scripts/workflow/jira-comment.py (comment body).
Boundary markers: No explicit delimiters or instructions are used to separate external content from the agent's internal logic.
Capability inventory: The agent can perform significant write operations including creating/updating issues, transitioning statuses, logging work, and downloading files.
Sanitization: The skill extracts text from Atlassian Document Format (ADF) in scripts/lib/output.py but does not perform semantic sanitization to detect or neutralize instructions.

jira-communication