slack-feedback-to-issues
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
glabCLI to search for and create issues. It interpolates data derived from Slack messages (such as bug report terms, titles, and descriptions) directly into shell commands. If the Slack content contains shell metacharacters (e.g., backticks, semicolons) and the agent does not properly escape them, this could lead to unintended command execution. - [INDIRECT_PROMPT_INJECTION]: The skill regularly ingests untrusted data from a Slack channel to automate issue creation.
- Ingestion points: Slack messages and thread replies (SKILL.md, Procedure step 1).
- Boundary markers: The procedure includes a mandatory confirmation step (Procedure step 4) where the user must approve the summary, title, and labels before the issue is created.
- Capability inventory: The skill can search and list issues, view issue details, create new issues, and list project labels using
glab. It also writes state totmp/slack-feedback-state.json. - Sanitization: The instructions do not explicitly mandate shell-escaping or input sanitization, although the 'Traps' section advises summarizing rather than copying private data for public projects.
Audit Metadata