Research

Technical work on policy-gated agent execution, human authorization, provenance, attestation, and reconciliation. We publish working papers, experiments, architecture notes, and implementation reports — including limitations and negative results.

Type

All Architecture note Experiment Implementation report

Topic

All Policy and consequential-action classification Connector and enforcement design Prompt-injection containment Attestation and provenance Human authorization Execution reconciliation Agent assurance and certification

1 publication · Clear filters

Experiment10 August 2026

Containing untrusted CI output with a quarantined reader

CI job logs are third-party-influenced text that flowed unfiltered into a privileged planning prompt. We routed them through a zero-tool-access reader restricted to schema-validated output, and tested it against a real injection.

Caretaker ResearchPrompt-injection containment Attestation and provenance