enterprise-compliance

Pulling AI agents into an enterprise compliance program has often meant stitching together ad-hoc scripts and spreadsheets. The enterprise-compliance plugin adds a ready-to-use compliance layer on top of DeepSeek Harness (DSH), spanning SOC 2 and GDPR automated self-assessment, sensitive-data protection, and full operation auditing, so compliance becomes a continuously observable capability rather than a quarterly scramble.

The plugin closes the loop with seven model tools and three quick-win enhancements. compliance_report scores each control against CC 6.1, CC 7.2, GDPR Article 5 and more, recording a trend over time; compliance_redact applies nine regex patterns to scrub emails, national IDs, private keys, API keys and similar fields from telemetry before export; compliance_audit persists every tool invocation with timestamp, result, session and redacted parameters into settings.yaml so evidence survives restarts, with multi-format export and filtering by tool, session or time window. Additional capabilities include workspace file scanning for plaintext secrets respecting .gitignore by default, GDPR Article 20 data export and Article 17 one-click erasure, and threshold-based alerts that fire a host event and, when configured, POST a JSON payload to a Webhook endpoint. Every policy knob covers check toggles, alert threshold, audit retention, redaction rules and alert Webhook, all adjustable live from the Enterprise Compliance Center panel or via the compliance_policy tool with immediate persistence.

The plugin depends only on DSH official contracts and installs cleanly on both web and desktop profiles with no extra backend services. It is well suited to SaaS teams, internal engineering platforms, and DevOps toolchains that embed AI agents in production workflows, particularly engineering leads and platform architects who need to demonstrate to auditors that AI invocation chains are traceable and that sensitive data has been minimized.