What it is
jhf-harness is the open-source governance runtime for controlled AI-agent execution inside the Helpifyr stack. It is the layer between a proposed action and actual execution.
Governance / Runtime
jhf-harness
Governance runtime for controlled AI-agent execution.
jhf-harness checks planned AI-agent actions against roles, policies, approvals, evidence, and compliance rules before they are executed.
Open Autonomy Spine
Status Work in progress
Status
Status: Work in progress / not production ready
- experimental and not yet intended as a production compliance, security, or decision-authority layer
- evaluates, allows, blocks, or escalates - but does not make the business decision alone
What it does
jhf-harness checks actions before execution.
It classifies risk, requests approvals, and emits evidence.
It writes outcomes back into Helpifyr in a controlled way.
Scope v1
The first version stays intentionally bounded. It shows the governance surfaces that sit between an agent proposal and Helpifyr writeback.
At the core
Action Proposal. An agent proposes a concrete action with target, context, and expected effect.
Case Bundle. The runtime assembles roles, data sources, evidence, and relevant control information.
Policy Check. Policies and boundaries are checked against the case before execution.
Risk Classification. The action is classified as low-risk, approval-required, or blocked.
Human Approval Gate. Critical steps are escalated to accountable humans instead of being executed silently.
Output Contract. Only allowed, controlled outcomes are admitted into the next step.
Evidence Event. Decision, source, and approval remain available as reviewable evidence.
Writeback in Helpifyr. The outcome is written back into the Helpifyr process in a controlled way.
How it sits in the stack
jhf-harness sits between orchestration, business truth, compliance rules, and human approval.
- Warp orchestrates agent and workflow execution
- Spindle remains the business truth and target system for controlled writeback
- Selvage supplies the relevant compliance rule profiles
- Lantern visualizes human approval and review without being pushed as a separate public brand in this wave
- Harness evaluates and gates the action before execution
v1 demo flow
The demo flow shows the minimum governance chain between an agent idea and controlled writeback.
Agent proposes an action
An agent formulates a planned action with target and context.
Harness builds the case bundle
Roles, sources, policies, and evidence are bundled for the decision case.
Harness classifies the risk
The runtime decides whether the action may proceed, must escalate, or must be blocked.
Selvage supplies the matching rule
The relevant compliance rules are supplied as reviewable runtime inputs.
Human approval is requested
Critical steps stay approval-gated instead of running autonomously.
The decision is recorded
Evidence, decision, and rationale are captured as reviewable records.
The outcome is written back into Helpifyr
Only the controlled outcome reaches the next operational step.
How it fits into the system
jhf-harness does not stand alone. It connects to neighboring modules so a single capability becomes dependable follow-through.
Important boundary
jhf-harness stays bounded to its role as Governance runtime for controlled AI-agent execution. It does not replace other modules; it makes its part of the system traceable, connectable, and reviewable.
What comes next
What it is not
The current scope stays intentionally narrower than a complete governance or compliance universe.
- not a complete compliance universe
- not its own legal logic
- not a certification
- not an automatic compliance guarantee
Open-source basis
These are the technologies carrying the module behind the scenes. They stay secondary, but they help technical readers orient themselves.
Why open governance infrastructure matters
AI agents are becoming increasingly action-capable. To use these systems responsibly in companies, SMEs, research, and public institutions, open, traceable, and reviewable governance building blocks are needed. jhf-harness is meant to make that control layer available as open-source infrastructure.
- public problem: the AI Act and governance-readiness gap for SMEs
- reuse value beyond Helpifyr for governance runtime, rule profiles, and control patterns
- reusable building blocks: governance runtime, compliance rule profiles, evidence/audit events, human-approval gates, and writeback patterns
Translated from Solarisara
Solarisara describes value, role, and decision principles for controlled autonomy. jhf-harness translates those principles into a technical governance runtime inside the Helpifyr stack without pushing the Solarisara storyworld to the foreground.
What keeps this page honest
This explanation stays anchored to the module’s current truth, including its real boundaries, responsibilities, and contracts.
jhf-harness is the technical governance runtime for controlled AI-agent execution in the Helpifyr stack.
solarisara/solarisara-autonomy-spine
README.md
Open-source truth
This page is rendered from the repo-owned projection truth and remains tied to the README, module boundaries, and status.
License note: AGPLv3. The current open reference source for this governance runtime lives in the Solarisara Autonomy Spine repository, from which the Helpifyr path is technically translated.
Open-source source http://192.168.1.166:3000/solarisara/solarisara-autonomy-spinejhf-harness
jhf-harness sits between orchestration, business truth, compliance rules, and human approval.