BitssyBytes logoAgent Safety Score

Directory / OpenHands

B

OpenHands

Coding agent (open source) · OpenHands · Last checked 2026-10-16

Runs actions in a Docker sandbox by default. Confirmation mode plus an LLM security analyzer can pause risky actions for a human.

Overall: B (2.00 avg over 4 verified criteria)

Lock this down first

Set SECURITY_CONFIRMATION_MODE=true. It is off unless you turn it on.

Scorecard

CriterionRatingNotes
Permission granularityUnverifiedNot confirmed in the docs we checked.
Human approval gates2/3Confirmation mode and a risk analyzer, configured by env or SDK.
Sandboxing2/3
Secrets handling2/3Secrets manager exports values as env vars. The SDK SecretRegistry masks secret values in tool output.
Audit logsUnverifiedNot confirmed in the docs we checked.
Kill switch / interrupt2/3Conversations and sandboxes can be paused, resumed or deleted via SDK and API.
Network egress controlUnverifiedNot confirmed in the docs we checked.

Unverified: Permission granularity, Audit logs, Network egress control. That means we could not confirm it in public docs, not that it is missing.

Sources