Integration
Claude Code
Add runtime policy in front of Claude Code hooks so low-risk reads stop spamming the user and high-impact actions can be reviewed or blocked.
Open integration guide5D integrations
Hook 5D into the runtime you already use.
Use 5D when your agent can write files, run shell, call external APIs, or touch sensitive tools.
Integration
LangGraph
Score the pending tool call, route by band, and only interrupt the graph when the action actually deserves review.
Open integration guideIntegration
PydanticAI
Use 5D as the decision function behind `ApprovalRequiredToolset` so the agent only defers tool calls that actually cross your risk boundary.
Open integration guideIntegration
OpenAI Agents SDK
Treat risky tool calls as tripwires and let 5D return a normalized `allow` / `review` / `deny` decision that your OpenAI runtime can act on.
Open integration guideIntegration
Claude Agent SDK
Put a runtime policy layer in front of repo edits, CI reruns, webhooks, and risky maintenance actions so AI coding agents can keep moving without getting reckless.
Open integration guideTry 5D
Start from the repo, then wire the adapter you need.
5D stays small on purpose: local install, configurable policy, provider-neutral runtime decisions, and a clean review handoff when the action deserves it.