AI Product Manager · AI Coding / Developer Tools · Business AI Agents
I turn ambiguous product goals into executable, verifiable agent workflows—from product discovery and acceptance criteria to delivery evidence and release operations.
Portfolio / 个人简历 · Open-source repositories
| Project | Built for | What it proves | Start here |
|---|---|---|---|
| open-workflow-kit | AI coding teams | Completion Contracts, evidence-led delivery and cross-tool governance | 30-second demo |
| business-agent | Business AI agent projects | Planning, gateway scaffolding, security boundaries and evaluation workflows | npm |
| open-data-agent | AI agents querying business data | Whitelisted SQL, L1-L4 access control, row-level isolation, deterministic masking/refill, MCP and auditability | 3-minute demo |
| openone-workflow-kit | Solo builders | Product delivery and commercialization in one verifiable workflow | 30-second demo |
AI coding team delivery → open-workflow-kit
Business AI agent lifecycle → business-agent
Secure business data access → open-data-agent
Solo product + growth → openone-workflow-kit
The three workflow kits share one principle: “done” should be backed by observable evidence, not a feeling. Open Data Agent applies the same evidence-first approach to business-data access through deterministic controls, explicit boundaries and auditability. Each repository states its tested boundaries explicitly; structural or contract tests are not presented as real-production certification.
- Reproducible CI, package and release workflows
- Tool adapters with clearly separated conformance checks and real-client verification boundaries
- Security, privacy and rollback rules designed into the delivery workflow
- Whitelist-based data access with server-side authorization, isolation, masking and audit trails
- Product decisions documented as acceptance criteria, evaluation paths and post-release evidence
- AI-native product delivery and developer tooling
- Business agent reliability, evaluation and operational readiness
- Secure, auditable business-data access for AI agents
- Turning product strategy into workflows that agents and humans can verify together
For experience, case studies and contact details, visit liangfei.dev.