security-genome-reactor
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@security-genome-reactor infer security phenotypes from our latest scan findings"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Security Genome Reactor
Evolve scan and policy configurations against the fleet's measured phenotype instead of its labels.
Infers latent security phenotypes from observed findings rather than operator labels, seeds competing scan/policy genomes per phenotype, evolves them under measured fitness, and keeps an explicit parent-before-child ancestry graph for every configuration in production.
It is an analysis and decision surface, not an actuator: it has no network client, touches no files, spawns no processes, and reads no environment variables.
Why a practitioner would install this
Tags lie; behaviour does not. Phenotypes are inferred from observed findings and compliance behaviour, so a mislabelled asset group stops driving the scan policy.
Configurations compete instead of accumulating. Each phenotype gets several candidate genomes that are scored, not one inherited profile nobody dares change.
Fitness can be measured, not asserted. Callers may supply real measured fitness per genome; rehearsal is the fallback, never a substitute presented as measurement.
Every configuration has parents. Ancestry is returned in parent-before-child order, so 'why is this profile like this' has an answer.
A genome becomes a bounded trial. The reactor emits an explicit target/policy/profile trial plan rather than mutating anything.
Related MCP server: genetic-mcp
Behavioural contract
genome_infer_phenotypesinfers phenotypes from records above an explicit threshold and ignores cloud labels as ground truth.genome_seedseeds multiple competing genomes for one phenotype.genome_evolveruns one deterministic generation using caller-supplied measured fitness keyed by genome id, or rehearsal when none is supplied.genome_trial_plantranslates one genome into a bounded target/policy/profile trial plan.genome_lineagereturns ancestry in parent-before-child order and rejects cycles.Populations below two genomes, unknown ids and invalid thresholds fail closed.
Prerequisites
Node.js 20 or newer (
node --version). Zero runtime dependencies.An MCP client that speaks stdio (Claude Code, Claude Desktop, Cursor), or direct library use from TypeScript.
No API key, account, network access or Tenable product is required.
Install and run
git clone https://github.com/SweetKenneth/shpbl-security-genome-reactor.git
cd shpbl-security-genome-reactor
npm install # devDependencies only: typescript
npm run build # compiles to dist/
npm test # 39 behavioural, boundary and fail-closed tests
npm start # starts the MCP server on stdioMCP client configuration:
{
"mcpServers": {
"security-genome-reactor": {
"command": "node",
"args": ["/absolute/path/to/shpbl-security-genome-reactor/dist/src/mcp-server.js"]
}
}
}Tools exposed
genome_infer_phenotypes— Infer latent security phenotypes from observed findings and compliance behavior, ignoring cloud labels as ground truth.genome_seed— Seed multiple candidate security genomes for one phenotype.genome_evolve— Run one deterministic evolutionary generation using rehearsal by default or caller-supplied measured fitness keyed by genome id.genome_trial_plan— Translate a genome into a bounded target/policy/profile trial plan.genome_lineage— Return ancestry in parent-before-child order for a genome from a supplied population/history.
What it outputs
Inferred phenotypes with supporting evidence, competing genome populations, per-generation fitness and selection results, bounded trial plans, and ordered ancestry graphs.
Verification
Reproduce all of it from a clean clone with npm run check:
Strict TypeScript compile and
--noEmittypecheck: PASSBehavioural tests: 39/39 PASS
Randomised invariant hammer: 30,000 cases / 420,000 invariant checks PASS
Static scan for network, filesystem, process and dynamic-eval surfaces in
src/: PASS (0 findings)Worked example runs end to end: PASS
Runtime dependencies: 0
Known limitations
No scanner or cloud client is embedded; records, fitness and execution are supplied by the caller.
Rehearsal fitness is an explicit fallback and is labelled as such — it is not a measurement of your fleet.
Phenotype inference is a deterministic clustering of supplied observations, not a claim about intent or ownership.
The reactor proposes trials. It never changes a scan, policy or profile.
Provenance and lineage
This product exists because two things were put together, and both are credited.
Upstream capability inspiration — conard0-git/targeted-nessus-scan, by Isaac Conard (conard0-git), MIT licensed. Its observed behaviour was studied as a capability surface: what a practitioner in that domain actually needs to do. The exact paths and lines that were read are recorded in PROVENANCE.json. No line of upstream implementation code is used in this package. The upstream licence text is preserved under THIRD_PARTY_NOTICES/ as provenance; it does not license this implementation.
SHPBL capability library — shpbl.com. SHPBL (shpbl.com) is a governed library of reusable software capabilities and a method for composing them: it reads a target repository, identifies what capability it demonstrates, matches that against owned capability records, and writes new software where neither side had it before. The capability parents used here are listed by identifier in PROVENANCE.json. No harvested capability body is embedded in this package.
The implementation in this repository was written fresh from the approved capability contract for this run. The literal composition is 0% upstream code, 0% copied SHPBL capability bodies, 100% new implementation. That is an exact-line and byte-level statement about this source tree, not a legal opinion.
Author and copyright: Kenneth E. Sweet Jr., MIT licensed.
Attribution does not imply endorsement by Isaac Conard (conard0-git), Tenable, or any other party.
Tenable status
Submitted to the Tenable CyberAgents Exchange for review on September 14, 2026 — pull request #173. Submission does not imply review, approval, certification, validation, endorsement or acceptance by Tenable.
Files
src/— implementation and the stdio MCP server.tests/— behavioural, fail-closed and MCP integration tests.scripts/— randomised invariant hammer and the static security scan.examples/worked-example.ts— an end-to-end run you can execute.SECURITY.md— threat boundary and forbidden behaviour.PROVENANCE.json— upstream and SHPBL capability lineage.MANIFEST.json/CHECKSUMS.sha256— released file inventory and hashes.LICENSE— MIT.
License
MIT © 2026 Kenneth E. Sweet Jr.. See LICENSE.
Available Tools
5 toolsgenome_evolveB
Run one deterministic evolutionary generation using rehearsal by default or caller-supplied measured fitness keyed by genome id.
Inputs: phenotype (required), population (required), fitnessByGenomeId.
genome_infer_phenotypesA
Infer latent security phenotypes from observed findings and compliance behavior, ignoring cloud labels as ground truth.
Inputs: records (required), threshold.
genome_lineageB
Return ancestry in parent-before-child order for a genome from a supplied population/history.
Inputs: genomes (required), genomeId (required).
genome_seedC
Seed multiple candidate security genomes for one phenotype.
Inputs: phenotype (required).
genome_trial_planC
Translate a genome into a bounded target/policy/profile trial plan.
Inputs: genome (required), phenotype (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool targets a clearly distinct step in the genome workflow: inferring phenotypes, seeding candidates, evolving generations, planning trials, and tracing lineage. There is no meaningful functional overlap between any two tools.
All names share the genome_ prefix and use snake_case, which is readable and recognizable. However, the pattern is mixed: genome_infer_phenotypes is verb-object, genome_seed and genome_evolve are bare verbs, while genome_trial_plan and genome_lineage are noun phrases.
Five tools is well-scoped for this specialized domain. Each tool represents a meaningful stage in the reactor lifecycle without redundant or filler operations.
The set covers the core genome lifecycle: infer phenotypes, seed candidates, evolve generations, produce trial plans, and inspect lineage. Fitness is intentionally caller-supplied, so no evaluation tool is a necessary gap.
Related MCP Connectors
Developmental agents: create an agent from a digital genome, then evolve it by evaluation.
Threat modeling, code/cloud/pipeline scanning, shadow-AI discovery, compliance checks and fixes.
Autonomous AppSec & DAST scanner across 12 security domains (260+ controls) with 3-Layer Triage.
Discover software problems, analyze evidence, and create implementation-ready Build plans.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables evolutionary solution generation and optimization using genetic algorithm principles, allowing LLMs to evolve solutions across multiple generations with consistency check evaluations and convergence detection.5MIT
- AlicenseNot gradedqualityDmaintenanceEnables evolutionary idea generation using genetic algorithms with LLM workers, multi-objective fitness evaluation, and advanced genetic operations.1MIT
- FlicenseNot gradedqualityDmaintenanceMCP-native evolution control plane that discovers, tests, governs, versions, and promotes improved AI agents. Integrates with LocalLLM-MCP for real or simulated model evaluation.-
- AlicenseAqualityBmaintenanceEnables agents and security automation to adjudicate proposed defense mutations against a sealed, canonicalized attack scenario, enforcing reproduction, impact-screening, regression, and proven-improvement gates before promoting at most one candidate. Every verdict, rejected candidate, and rejection reason is sealed into a recomputable SHA-256 Merkle evidence root and hash-linked lineage that can be independently verified without trusting the server.6MIT