Security audit
Security checks for vulnerabilities and agentic risk
The SendGrid workflow has appropriate confirmation steps, but its fallback setup tells the agent to run downloaded installation code without asking first.
Review and approve CLI installation separately before using the missing-dependency fallback, including its automatic skill synchronization. Use a verified release and narrowly scoped SendGrid credentials, and retain confirmation before sending email. The Review verdict concerns setup control and unverified downstream installation effects; the supplied artifact does not establish malicious behavior.
The supplied command installs software from an external CLI provider; it does not show any SendGrid operations or establish that this installation supports them. Executing a remote installer is a materially different capability from the declared purpose. The downloaded script's contents and further effects are not provided and cannot be determined from this chunk.
Piping a live download directly into Bash executes whatever the installation endpoint returns, without a pinned version, independent integrity check, or review; the PowerShell alternative has the same weakness. If the endpoint or distribution pipeline is compromised, execution could expose local files and credentials or modify the host with the agent's privileges, although this path is conditional and may be blocked by the declared tool allowlist.
oo: command not found — install the oo CLI (other platforms: https://cli.oomol.com/install-guide.md):
curl -fsSL https://cli.oomol.com/install.sh | bash # macOS / Linux
The fallback directs the agent to install software through remotely supplied scripts without requiring approval for local code execution or explaining potential system changes. Although conditional on a missing dependency, it extends a SendGrid task into local software installation; strict enforcement of the declared Bash(oo *) permission would limit execution of the shown installer commands.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.
- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.
## Resources