Managed API
A live runtime behind https://api.hrevn.com for baseline checks, bundle generation, verification and bundle download.
HREVN
Technical Overview
Live runtime, public surfaces, and verifiable continuity for AI workflows
HREVN is no longer only a protocol hypothesis. It now operates as a live managed runtime with public surfaces for Anthropic, Codex, Google / Genkit, OpenClaw, and Kiro, plus a public MCP server, a public npm package for the Google surface, and a public Kiro power alpha.
View developer entry points Open developer layer Managed API health
What HREVN is
HREVN is verifiable workflow infrastructure for AI agents. Its role is not to replace the model layer or the orchestration layer, but to add a durable continuity layer around consequential execution: baseline diagnostics before important steps, verifiable evidence after execution, and structured outputs that make interruption and resumption more reliable.
In practical terms, HREVN helps answer three questions that ordinary logs and chat history answer poorly: what happened, what remains missing, and from which trusted point a workflow can safely continue.
Current Runtime Shape
MCP Server
A public MCP server published through the MCP Registry and PyPI for structured tool access to the same runtime.
Public Surfaces
Thin public surfaces for Anthropic, Codex, Google / Genkit, OpenClaw and Kiro, all pointing back to the same operational truth.
Why continuity matters
Agent workflows often fail in ambiguous ways. A long execution can stop after partial work, but the surrounding system may still struggle to tell what completed, what failed, or which checks were never reached. Without a verifiable continuity layer, teams rebuild context from memory, logs, or chat transcripts, repeating work and losing confidence.
HREVN addresses this by returning structured outputs such as profile detection, readiness level, missing required blocks, risk flags, recommended next step and a remedy payload. That remedy path is important: the runtime does not only issue a verdict, it also explains what evidence is still missing.
Governance and evidence
HREVN does not make a system legally compliant on its own. What it does provide is a structured evidence layer that governance, audit and compliance processes can use: human oversight references, risk documentation, evidence lifecycle hooks, technical documentation references, receipts and bundle verification.
This makes HREVN particularly relevant in AI governance discussions where the challenge is not only model quality, but also whether a system can later prove what it did, under which authority, and with what missing controls still visible.
Three pages for the conceptual core
The technical overview explains the full system. These three pages frame the same product through continuity, evidence and verifiable state.
Workflow continuity
The operational angle: how to continue from the last trusted point.
Audit-ready AI workflows
The evidence angle: which outputs help review, audit and governance.
Verifiable workflow state
The category angle: why HREVN speaks about verifiable state rather than only logs.
Public entry points
Anthropic
Claude Code-facing surface using skills, a local runner and an MCP path.
Codex
Helper-first technical alpha for verifiable workflow state in coding-oriented agent flows.
Google / Genkit
Thin wrapper plus public npm package for the live HREVN runtime.
OpenClaw
Agent-first, local-first surface with a public PyPI CLI and real end-to-end baseline validation.
Kiro
Public power alpha for a spec-driven path with MCP, steering, and example hooks. Current status: honest exploration, not a finished native integration.
Current status
HREVN is currently in technical alpha across its public surfaces. That means the runtime is live, the first test paths are real, and a technical user can already discover, install and run the public artifacts. It does not yet mean fully productized onboarding, marketplace-grade packaging across every surface, or final enterprise distribution.