The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of subprocess.run to invoke the acli (Atlassian CLI), gog (Google Workspace CLI), and git commands. These are used to fetch Jira issue data, Google Sheet contents, and repository configuration.
Evidence in scripts/release.py, scripts/_jira.py, and scripts/rich_filter.py shows commands like acli jira workitem search and gog sheets get being executed with arguments derived from skill logic.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Jira issue descriptions, summaries, and Google Sheets to generate reports and Slack messages.
Ingestion points: scripts/release.py fetches data from Google Sheets (TEAM_SHEET_ID, SCHEDULE_SHEET_ID) and Jira issues via acli.
Boundary markers: The skill uses triple-backtick blocks (e.g., ```slack) to delimit generated output, but lacks explicit instructions to ignore embedded commands within the ingested data.
Capability inventory: The skill has the ability to read from the network (via CLI tools) and write to the terminal/agent context. It does not perform direct file writes or arbitrary system modifications.
Sanitization: The script _adf_to_text in scripts/_jira.py flattens Atlassian Document Format (ADF) into plain text, which removes some formatting but does not filter for malicious prompt instructions embedded in issue descriptions.
[EXTERNAL_DOWNLOADS]: The skill depends on external CLI tools (acli, gog, uv) and fetches data from remote APIs (issues.redhat.com, docs.google.com). These are well-known services consistent with the developer workflow of the 'redhat-developer' vendor.

rhdh-release-teams