Agent detail

Computer Agent

Natural-language computer interface capable of coding, terminal execution, and desktop workflows.

Audit-first trust modelProvenance awareEnterprise policy controls

VerifiedPassedLow Risk87ScoreSource VerifiedSecurity FailedScanned 29 Mar 2026

Description

What this agent does and how it is scoped.

Natural-language computer interface capable of coding, terminal execution, and desktop workflows. Published and verified by TrustAgent.

audit-firstcomputer-usedesktopenterprise-readyorchestrationterminaltrustagent-published

Trust Snapshot

Trust badges

Granular trust badges with evidence-driven status.

Manual Review PassedPublisher VerifiedEnterprise SafeBehavior VerifiedSandbox TestedPermission MatchedDependency Clean

Provenance

Source type

GITHUB

Publisher

TrustAgent

Source license

Apache-2.0

Commit hash

681f5ce5b84bc96a2a4cc5e90daa6328f3f796e0

Version hash

2145b4fb2983d3b957394313b597328a9306053e81b8736e0c5cbdcde6483461

Audit Evidence

Version History

Hash-locked versions, source metadata, and drift context.

Version	Commit	Hash	Source	Created
1.0.0	681f5ce5b84bc96a2a4cc5e90daa6328f3f796e0	2145b4fb2983d3b9	TrustAgent	3/29/2026, 8:33:54 PM

Reviews

Brianna BuyerVerified install

Great trust signals and clear audit evidence.

Public trust report

Computer Agent completed the Trust Agent automated review pipeline for agent listings. Stage 1 configuration integrity scored 84/100 and stage 2 behaviour testing scored 85/100. The companion security scan produced 87/100 with outcome PASSED. Overall trust score: 87/100. Primary analyst watchpoints: Process execution appears under-declared, Network behavior lacks explicit egress guardrails. This public narrative excludes prompts, code, and protected configuration while preserving the buyer-facing trust rationale.