Skip to content

For Navy SBIR Reviewers

Preliminary, surrogate-based - not government-validated

The metrics below are measured on the synthetic, unclassified surrogate and are preliminary; they are not validated against government TMPC code and not validated by the Navy.

This repo lets you run the modernization pipeline yourself, with no API keys and no network, and inspect both the legacy input and the modernized output.

bash git clone https://github.com/577Industries/forge-evolve-tmpc cd forge-evolve-tmpc make demo # offline, deterministic, ~minutes

make demo runs the full pipeline on the synthetic surrogate and writes evidence to results/:

Phase-I ask (topic) What the demo does Evidence produced
1. Analyze architecture, identify modernization needs Roslyn/Tree-sitter discovery → dependency graph, complexity, business rules, crypto inventory results/discovery-report.json, results/business-rules.ttl
2. Conceptual AI models for analysis/refactor/cyber CLAR lift → risk-scored microservice decomposition → multi-agent transform results/migration-plan.json, results/clar/*.jsonld
3. PoC AI refactoring on a representative component Modernize a coupled C# component to .NET 8; prove behavioral equivalence results/equivalence-report.json, tmpc-modern-mds/
4. Phase II prototype plans + cyber cATO artifacts as CI byproducts results/stig-findings.*.json, results/control-map.yaml, results/sbom.cdx.json, results/provenance.json, results/poam.csv

What to read first: companion/companion.md maps each proposal claim to the exact file/test that backs it. EXCLUSIONS.md states why this is safe to publish.

Honesty note. Every metric the demo prints is measured on the synthetic, unclassified surrogate in surrogate/ and is preliminary - not validated against government TMPC code and not validated by the Navy. The surrogate is engineered to exhibit the same classes of technical debt and shapes of mission-planning logic the government described, so results transfer in kind, not in fidelity. See surrogate/DEBT.md.