teams-dev
Warn
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to install a global CLI tool using a remote binary package from the author's GitHub repository releases (
npm install -g https://github.com/heyitsaamir/teamscli/releases/latest/download/teamscli.tgz). It also includes ateams self-updatecommand that performs unverified remote updates of the CLI tool. - [EXTERNAL_DOWNLOADS]: Fetches software components from the author's GitHub repository and references external services such as Microsoft devtunnels and ngrok for exposing local endpoints to the internet.
- [COMMAND_EXECUTION]: Relies heavily on shell command execution for project scaffolding (
teams project new), app registration (teams app create), and Azure resource configuration using theazCLI. - [PROMPT_INJECTION]: The skill scaffolds bot projects using templates (e.g.,
ai,graph,mcp) which may contain instructions that could influence agent behavior during code generation or deployment. - Ingestion points: Project templates processed via the
teams project newcommand inreferences/guide-create-bot-app.md. - Boundary markers: Absent; no explicit delimiters are provided to isolate template content from agent instructions.
- Capability inventory: The skill has capabilities for file system writes, subprocess execution (
npm,pip,dotnet), and network access via the Teams and Azure CLIs. - Sanitization: Absent; there is no evidence of validation or sanitization applied to template content before it is written to the user's disk.
Audit Metadata