v0.4.0 alpha — latest published release

v0.4.1 hardening candidate: paid sales are disabled until license fulfillment and activation are verified end to end.

Auditable AI Actions

Cryptographic audit trails for AI agent workflows. Every action HMAC-SHA256 signed and hash-chained. Tamper with one step — the entire chain breaks.

$ pip install https://github.com/jdhart81/quantaoptima/releases/download/v0.4.0/quantaoptima-0.4.0-py3-none-any.whl

Existing Pro licenses require reissuance and a trusted public key. Read the migration guide.

GitHub Pro preview status
10
MCP Tools
SHA-256
HMAC Signed
0 deps*
Audit Core
30s
Setup Time

*Audit chain requires only Python stdlib. Optimizer demo uses numpy/scipy.

The Problem

AI agents are making thousands of decisions per day — querying databases, calling APIs, transforming data, optimizing parameters, writing code. But there's no audit trail. No way to prove what happened, when, or why.

For regulated industries, scientific research, and any team that needs accountability — this is a blocker. QuantaOptima fixes it.

How the Audit Chain Works

Action 1: "query" Action 2: "decide" Action 3: "execute" ┌──────────────────┐ ┌──────────────────┐ ┌──────────────────┐ │ before: {input} │ │ before: {options} │ │ before: {plan} │ │ after: {result} │ │ after: {choice} │ │ after: {output} │ │ sig: HMAC-SHA256( │──chain──│ sig: HMAC-SHA256( │──chain──│ sig: HMAC-SHA256( │ │ prev_sig + data│ │ prev_sig + data│ │ prev_sig + data│ │ ) │ │ ) │ │ ) │ └──────────────────┘ └──────────────────┘ └──────────────────┘ Tamper with Block 1 → Blocks 2 and 3 signatures become INVALID

Same principle as blockchain, applied to AI agent actions. Every block's signature depends on the previous block's signature. Change anything, anywhere, and verification fails from that point forward.

Two Ways to Use QuantaOptima

As an MCP Server

Your AI agent can log actions, verify the chain, and export the audit trail through natural language. 10 tools available out of the box. Setup takes 30 seconds.

{
  "mcpServers": {
    "quantaoptima": {
      "command": "quantaoptima-server"
    }
  }
}

As a Python Library

Embed the audit chain in your own MCP server or application. Use the @auditable decorator to auto-log any function call.

from quantaoptima import AuditChain

chain = AuditChain(scope="my-server")
chain.log("query", {"q": "..."}, {"result": "..."})
assert chain.verify()

10 MCP Tools

Core: Audit Chain (Free)

quantaoptima_log_action
Log any action with before/after state to the cryptographic chain
quantaoptima_verify_chain
Verify HMAC-SHA256 chain integrity — detect any tampering
quantaoptima_export_chain
Export the full audit trail as JSON for compliance
quantaoptima_chain_status
View chain statistics, action breakdown, and health

Demo: Quantum-Inspired Optimizer

quantaoptima_optimize
Run optimization — every step audited automatically
quantaoptima_explain
Human-readable explanation of optimizer decisions
quantaoptima_benchmark
Compare against scipy's classical methods PRO
quantaoptima_observe
Inspect entropy, interference, phase transitions PRO
quantaoptima_audit
Verify optimizer's dedicated audit trail PRO
quantaoptima_status
License tier, limits, and subscription management

Who Needs This

Regulated Industries

Pharma, finance, aerospace — use the HMAC chain to authenticate explicitly recorded workflow data. Independent controls are still required for identity, completeness, execution, and compliance.

Scientific Research

Reproducible AI workflows. Every data transformation, model call, and analysis step is cryptographically logged. Share the audit trail with your paper.

AI Governance Teams

Building trust in autonomous agents. The audit chain detects modification of explicitly logged inputs, metadata, and outputs; it does not prove omitted actions or caller-supplied identity.

Pricing

Community

Free
  • Full audit chain (unlimited)
  • Log, verify, export actions
  • Python library + decorator
  • Optimizer: 3 objectives, 10D
  • Community support
Install Free

Pro

$29/month
  • Everything in Community
  • Advanced chain analytics
  • All 6 optimizer objectives
  • 100 dimensions, 5000 iterations
  • Benchmark, observe, audit tools
  • Email support
Preview — not for sale

Public checkout is disabled pending fulfillment verification.

Enterprise

Custom
  • Everything in Pro
  • Reviewed built-in objectives only
  • Higher preview limits
  • Roadmap, not a current offer
  • No active SLA
Preview status

Honest Positioning

The built-in optimizer is a demo of the audit chain in action, not a replacement for scipy. On raw solution quality, scipy's Dual Annealing finds better answers at similar evaluation budgets. But scipy can't talk to your AI agent, can't prove its work cryptographically, and can't show you how it explored the landscape. Different tools for different problems.

Use QuantaOptima when you need auditable, interpretable AI agent actions. Use scipy when you only need a number.