20+ tools built · herndon, va

Deep intelligence
for modern networks.

Reasoning, data, and action, each governed behind deterministic policy and orchestrated from one console. Open source, built for depth.

// kraken@stargraph — stargraph serve live
// products

One governance plane, tool by tool.

Fathom [ OPEN SOURCE ]
v0.7.4 · reasoning

Python-first expert system runtime built on CLIPS. Deterministic reasoning with persistent working memory, compliance rule packs, and sub-millisecond latency.

// example
engine = fathom.load("rules/governance.yaml")

result = engine.evaluate({
  "agent": "analyst",
  "action": "access",
  "data_classification": "SECRET"
})
# => Decision(action="deny", reason="clearance insufficient")
Nautilus [ OPEN SOURCE ]
v0.1.5 · data governance

Policy-first data broker for AI agents. One request plans, routes, enforces, attests, and audits, with cumulative exposure tracked across a session.

// example
from nautilus import Broker

broker = Broker.from_config("nautilus.yaml")
response = broker.request(
    "agent-alpha",
    "Find vulnerabilities for CVE-2026-1234",
    {"clearance": "unclassified",
     "purpose": "threat-analysis",
     "session_id": "s1"},
)
# plans → routes → enforces → attests → audits
Railyard [ PREVIEW ]
v0.1 · orchestration

Full-stack platform for building and governing multi-agent systems. Go backend, React dashboard, DSPy reasoning modules, and CLIPS-based policy governors.

// example
workflow.stage("analyze", {
  module: "chain_of_thought",
  agents: ["analyst"],
  governor: "nist-800-53"
})
workflow.stage("review", {
  require_approval: true,
  on_failure: "escalate"
})
Stargraph [ OPEN SOURCE ]
v0.5.3 · orchestration

The open-source runtime Railyard is built on. A stateful agent-graph framework that composes LLMs, classical ML, and tools into auditable, replayable graphs where deterministic CLIPS rules decide every transition instead of a model.

// example
nodes:
  - id: classify
    kind: stargraph.nodes.dspy:Classify
rules:
  - id: route-low-confidence
    when: "(node-id (id classify)) (state (confidence ?c&:(< ?c 0.8)))"
    then: [{ kind: goto, target: human_review }]
  - id: route-high-confidence
    when: "(node-id (id classify)) (state (confidence ?c&:(>= ?c 0.8)))"
    then: [{ kind: goto, target: apply }]
Bosun [ DESIGN PHASE ]
· action governance

Agent action governance mesh. It controls what agents do, passing every action through Fathom-powered policy before execution.

// example
bosun.authorize({
  agent: "classifier",
  action: "escalate",
  context: ["session-47", "agent-a:output"]
})
# => Authorization(allowed=true, attestation="eyJ...")
// architecture · the descent surface → bedrock

A layered stack you descend, not a platform you buy.

Each layer owns one concern: reasoning, data, action, orchestration. Intent enters at the surface and sinks through policy until it reaches deterministic bedrock. Adopt any depth; the floor holds.

0 m
epipelagic
1 BAR · PRESSURE
TEMP20 °C
LIGHT100%
[ SUNLIT ]
AI Agents inputs

Untrusted agent intents enter the column here. Nothing is assumed about them.

200 m
mesopelagic
21 BAR · PRESSURE
TEMP12 °C
LIGHT2%
[ TWILIGHT ]
Stargraph orchestration
v0.5.3 · open source

Runs the agent graphs. It composes LLM calls, classical ML, and tools, with deterministic rules deciding every transition instead of a model.

deep dive →
1 000 m
bathypelagic
101 BAR · PRESSURE
TEMP4 °C
LIGHT0%
[ MIDNIGHT ]
Bosun action governance
design phase

Every action clears policy before it runs, with context declarations that build the audit trail.

deep dive →
2 000 m
abyssal
201 BAR · PRESSURE
TEMP3 °C
LIGHT0 lux
[ ABYSSAL ]
Nautilus data governance
v0.1.5 · open source

Mediates every read and tracks cumulative exposure so no agent quietly builds a dossier.

deep dive →
4 000 m
abyssopelagic
401 BAR · PRESSURE
TEMP2 °C
LIGHT0 lux
[ ABYSS-DEEP ]
Fathom reasoning
v0.7.4 · open source

Deterministic CLIPS decisions with sub-millisecond latency and a full audit trail.

deep dive →
6 000 m
hadal
601 BAR · PRESSURE
TEMP2 °C
LIGHT0 lux
[ HADAL-TRENCH ]
CLIPS Engine foundation

A battle-tested expert-system runtime. The bedrock everything above rests on.

// servicenow suite

The “+” suite for ServiceNow.

view all applications →

The same deterministic governance, scoped to the ServiceNow platform. Each app brings reasoning, data, and action policy to the instance, modeled like the open-source stack.

DMM Data Management+

Executive tool for managing Data Management Plans across the data asset lifecycle. It tracks the commitments attached to each plan and turns breaches of data obligations into operational alerts.

x_krn_dmm
DocP Document+

Controlled-document management and publishing for the suite. Documents move through versioned revisions, each driven by a configurable approval workflow, with references, CRMs, summary sheets, and templates around them.

x_krn_document
Comp+ Compliance+

Self-service compliance portal for waivers, exemptions, privileged access, and training and certification tracking. Compliance is scored per user, per asset, and per group, with close attention to expiry.

x_krn_comp
Audit+ Audit+

The assurance hub of the suite. It runs the audit lifecycle, plan and execute audits, capture findings, drive remediation, collect evidence, and record attestations, all mapped to NIST 800-53, NIST AI RMF, and DISA STIG frameworks.

x_krn_audit
GovP Governance+

Governance-board management. It organizes boards into a hierarchy, schedules and runs meetings, manages agenda items, and records decisions and follow-up action items, with WebEx integration.

x_krn_govp
Mixer+ Mixer+

A duration and SLA analysis engine. It measures how long ServiceNow records take against a commitment threshold, with configurable time windows and exclusions, and reports pass or fail.

x_krn_mixer
HierP Hierarchy+

The shared backbone of the suite. A configurable hierarchy and rollup engine that renders any app's structure as a tree, scores each node with figures of merit, and aggregates values from children to parents.

x_krn_hp
// latest · field notes view all posts →
2026.06.17 · 37 min Air-Gapped Agents: A Deterministic, Rules-Routed Architecture for DoD, Cleared, and Offline Workloads Whitepaper 2026.06.17 · 15 min Two Independent Stop Layers: Why One Budget Cap Isn't Enough for Autonomous Agents Blog 2026.04.14 · 10 min Policy-First Data Access: Why Your AI Agents Need a Broker nautilus