launchdarkly-approval-requests

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from approval request descriptions and reviewer comments via the get-approval-requests tool (SKILL.md). Boundary markers: The agent is explicitly forbidden from approving requests and must wait for a human decision in the LaunchDarkly UI. Capability inventory: The agent can modify feature flags and configurations using apply-approval-request (SKILL.md). Sanitization: The agent is instructed to verify reviewStatus is approved and status is pending before applying any changes.
  • [COMMAND_EXECUTION]: The skill instructs the agent to use platform-specific browser tools or system commands such as open or xdg-open to display approval request URLs to the user (SKILL.md). This is used solely to facilitate the human-in-the-loop hand-off process.
  • [SAFE]: All external resources and dependencies, including the @launchdarkly/mcp-server, trace back to the official vendor infrastructure. The skill follows security best practices by enforcing human oversight for sensitive operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 05:50 AM