Use case · AI agent governance
Put a governance gate between your AI agent and its actions.
FoldEngine's Agent Governance Gateway intercepts a proposed AI tool call, evaluates readiness, blocks unsafe actions, and exports evidence — locally, before anything executes.
The problem
AI agents propose actions — file writes, API calls, shell commands, deployments. Most systems either execute them directly or fail without a governed review in between. There is rarely a recorded, evidence-backed decision about whether the action should have proceeded.
What FoldEngine adds
- 1Proposal
- 2Dry-run
- 3Gate
- 4ALLOW / BLOCK / REVIEW
- 5Evidence pack
- 6Memory
- 7Continuity record
The decision is explicit and bounded to evidence. FoldEngine does not execute the action; it governs the proposal and hands your system a decision it can act on.
Who this is for
Teams building with Claude tools, OpenAI function calling, LangChain, AutoGen, MCP-style workflows, local agents, and internal automation that want a governed review before tools execute.
What you get
- Local HTTP API for proposing actions and reading decisions.
- MCP server skeleton (production MCP service not implemented).
- PASR gate with reason codes (allow / block / review).
- Evidence pack per decision.
- Intervention memory across proposals.
- Local continuity record.
Boundaries
- No action execution — FoldEngine never runs the proposed action.
- No hosted production service.
- No authentication or TLS.
- No production MCP service.
- No Stability Ledger bridge.
See the Agent Gateway product page and the closed-loop platform story.
Where to go next
Use the Runtime Preview to inspect the governed runtime path, or open the explainers when you want a simpler walkthrough of consequence boundaries and evidence continuity.
Preview only. No live repo access. No autonomous execution. New demos will be added separately once they match the current product story.