Due Diligence Agents

Changes since v1.7.1

• v1.8.0: Agent prompt architecture — no-code customization, single safety floor, deterministic severity, provenance
• Address Copilot round-4 review (PR #202): getattr receiver scoping + config-panel XSS regression
• Address Copilot round-3 review (PR #202): describe --format doc + panel ordering
• Address Copilot round-2 review (PR #202): provenance, validate/runtime parity, sandbox false-positives
• Address Copilot review (PR #202): dd-config severity, AD-3a docs, provenance, fail-closed loader
• Harden test suite: default per-test timeout (prevent CI wedge on SDK hang)
• Add comprehensive XSS regression test across all report fields
• Polish: fail-closed validate, deterministic merge, count consistency, doc parity
• Fix import sort in test_customization_loader (CI ruff parity)
• Security + correctness fixes from adversarial release re-audit
• Resolve release-audit findings: doc drift + preview validation
• Resolve all UX audit findings + release polish
• Fix: wire dd-config/ markdown customizations into assembled prompts
• Fix: sanitize clause-summary article IDs (caught by live E2E)
• Update premium model profile to Opus 4.8
• Wave 4: polish + full documentation sync
• Wave 3: governance, security depth, intelligence, reporting UX
• Wave 2: markdown customization, profiles, agents CLI, provenance
• Wave 1: reusability backbone + dead-code injection fix
• Wave 0: non-removable safety floor + single severity authority
• Add Docker Hub description auto-sync in release workflow
• Update Homebrew formula to 1.7.1

Install

pip install dd-agents==1.8.0

Docker

docker pull zoharbabin/due-diligence-agents:1.8.0

What's Changed

• Agent prompt architecture: no-code customization, single safety floor, deterministic severity, provenance by @zoharbabin in https://github.com/zoharbabin/due-diligence-agents/pull/202

Full Changelog: https://github.com/zoharbabin/due-diligence-agents/compare/v1.7.1...v1.8.0