team-collaboration-standup-notes
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from multiple external sources.
- Ingestion points: The skill reads data from Git commit history, Jira ticket descriptions/comments, and Obsidian vault notes as specified in
resources/implementation-playbook.md. - Boundary markers: The AI prompt template provided in the implementation playbook does not use clear delimiters or instructions to ignore embedded commands within the processed data.
- Capability inventory: The skill is capable of summarizing data and, according to the provided reference scripts, posting that data to communication platforms like Slack.
- Sanitization: There is no evidence of filtering or sanitizing the content of commits or tickets before they are included in the prompt context.
- [DATA_EXFILTRATION]: The skill's primary purpose involves collecting and transmitting potentially sensitive internal data (project updates, commit logs, and meeting outcomes) from private repositories and task trackers to team communication channels. Users should ensure the target destinations (e.g., Slack channels) are appropriate for the sensitivity of the data being aggregated.
- [COMMAND_EXECUTION]: The
resources/implementation-playbook.mdfile provides a detailed Bash script (generate-standup.sh) and instructions for using various CLI tools such asjira,slack-cli,gcal, andclaude-ai. While these are legitimate developer tools, their execution involves piping data and potentially sensitive context through multiple third-party interfaces.
Audit Metadata