OpenHands (first-party docs)Verified

BL-CLM-49.

OpenHands — Human approval: SDK confirmation policies (always, never, risky) can require approval of actions; documented for the SDK, and one call path bypasses them.

Claim visualization

Actual claim data

BL-CLM-49 / Relationship

Claim evidence map

OpenHands (first-party docs)

OpenHands — Human approval: SDK confirmation policies (always, never, risky) can require approval of actions; documented for the SDK, and one call path bypasses them.

Documented for the SDK; the page says execute_tool() bypasses these controls. Not checked for Cloud/Canvas UI or defaults.

Source: docs.openhands.dev/sdk/guides/security ↗
Where in the source: Confirmation policy and Security analyzers sections
Source organization: OpenHands (first-party docs)
Evidence type: Vendor documentation
Source date: Not shown on the source
Last verified: October 13, 2026
Approved wording: According to OpenHands (first-party docs), SDK confirmation policies (always, never, risky) can require approval of actions; documented for the SDK, and one call path bypasses them.
Scope and caveats: Documented for the SDK; the page says execute_tool() bypasses these controls. Not checked for Cloud/Canvas UI or defaults.
Do not infer: Do not imply approval gates cover every agent action unless the source says so.

Cited onSelf-hosted AI agent platforms comparison ↗

← All claimsRecord BL-CLM-49