The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The scripts/install.sh file and SETUP.md instructions explicitly use the pattern curl -sSfL https://plannotator.ai/install.sh | sh. This executes unverified remote scripts with local shell privileges, which is a high-risk pattern for arbitrary code execution if the remote source or transmission is compromised.
[DATA_EXFILTRATION]: The scripts/phase-gate-review.sh script is designed to submit phase deliverables (PRDs, Architecture specs, etc.) to the external plannotator.ai service using the plannotator submit command. This represents a transfer of potentially sensitive project documentation to a third-party server.
[EXTERNAL_DOWNLOADS]: The skill initiates several external downloads and tool installations at runtime, including the plannotator CLI, fabric CLI, and potential project templates from the supercent-io GitHub organization. These dependencies are fetched from unverified external sources.
[COMMAND_EXECUTION]: Multiple scripts (scripts/init-project.sh, scripts/check-status.sh, scripts/phase-gate-review.sh) use shell commands to manage project state, manipulate files with sed, and execute Python snippets (python3 -c). These operations include the use of chmod +x on scripts and directory creation, which requires cautious management of path variables to prevent command injection.
[SAFE]: The skill follows standard configuration management patterns by using YAML templates for project settings and local files (docs/bmm-workflow-status.yaml) for state persistence. Documentation for integration with local notes applications like Obsidian and Bear correctly describes these as user-configured options.

bmad