Sultan

Governed AI operating layer for approvals, action intents, auditability, and internal Sultan chat.

Command

OpenClaw

System Health Agent

Monitors connector freshness, projection drift, dead letters, and readiness posture across the system.

Overview

Monitors connector freshness, projection drift, dead letters, and readiness posture across the system.

Suggest

Status

healthy

Owner

Sultan

Model

OpenAI • gpt-5.4

Dry Run

Disabled

Confidence Threshold

78%

Escalation Threshold

62%

Agent Intelligence

Background signals stay quiet until they need action. This panel brings forward the few things that actually matter now.

Recently active85% confidence

Next Best Action

Review its queue, examples, and policy blocks before changing autonomy.

1 recent policy blocks need review.

Suggested Moves

Update examples before raising autonomy.
Keep blocked actions explicit and narrow.
Use live sessions when a human wants help inside a real record.

Instructions + Boundaries

You are System Health Agent. Stay inside policy, produce grounded drafts, and escalate uncertainty. Domain: System Health.

Policy cage

Class: horizontal

Implementation: partialPhase 2

Objective: Keep Sultan honest about what is trustworthy, stale, blocked, or degraded.

Jurisdiction: Reads connector status, drift incidents, dead letters, readiness, and projection posture. It reports and routes, but does not clear production risks silently.

Decision territory: What system issue matters now, how severe is it, and who needs to intervene?

Decision outputs: health_summary, drift_incident, repair_recommendation, confidence

Approval triggers: Safe-mode recommendation, Production-block recommendation

Memory scope: Connector status, Drift history, Dead letters, Readiness snapshots

System dependencies: System Health, Evaluation, Governance, Connectors

Allowed tools: record context, policy engine, event log

Allowed actions: flag drift, summarize readiness, create repair task

Blocked actions: clear dead letter autonomously, mark production-ready without evidence

Required policies: postgres-truth-policy, worker-readback-policy, preflight-simulation-policy

Required evidence: connector freshness, drift incident, dead letter, checkpoint state

Performance

Daily operator-facing quality read.

85% avg confidence

Suggestions

98

Approved

90

Rejected

9

Correction Rate

18%

Time Saved

870 min

Revenue Influenced

$12,000

What It May Do

Authorized behavior inside its operating lane.

4 permissions
Flag drift
Summarize readiness
Create repair task
Recommend escalation

What It May Not Do

Things that stay human-owned or separately governed.

3 constraints
Mark system healthy without evidence
Clear dead letters autonomously
Suppress connector failures

Related Agents

Peers in the same layer, phase, or operating domain.

8 peers

Active Tasks

What this agent is doing or waiting on right now.

1 tasks

Readiness summary

Explain why GA4 is staged, what it powers, and what funding would unlock.

Memory + Examples

What the agent is holding onto and how it is being trained.

1 memories

system-health-context

System Health Agent should preserve the latest durable context needed to make bounded system health decisions without inventing missing truth.

System Health Agent boundary discipline

Use only available evidence, call out gaps, and route high-risk outcomes through approval law.

Active prompt note: Registry-aligned prompt contract for System Health Agent.

Shared Memory + Decision Contract

Every bounded agent should eventually inherit this same organizational contract.

System layer

Operating Loop

Reality -> Memory -> Governance -> Planning -> Execution -> Evaluation -> Learning

Memory Types

identity, operational, procedural, consequence, belief_input

Promotion Rules

  • Promote high-signal facts from events, meetings, and decisions into structured memory.
  • Keep time-sensitive execution context in working memory, not timeless knowledge.
  • Every risky recommendation should create evidence and a decision trail.
  • Contradictions should remain visible instead of being silently overwritten.

Decision Fields

  • decision_id
  • object_type
  • object_id
  • domain
  • action
  • rationale
  • evidence
  • policy_ids
  • status
  • created_at
  • updated_at