OPTIMIZATION & SWARMS DETERMINISTIC LOCAL AIR-GAPPED
SwarmDev Deterministic Pipeline
Turn natural language into inspectable, testable execution plans across 6 explicit stages with diagnostic sidecars.
SwarmDev Deterministic Pipeline
SwarmDev is a deterministic prompt-to-workflow compilation pipeline. It transforms unstructured natural-language requests into fully verifiable, inspectable execution plans backed by an immutable diagnostic trail.
6-Stage Deterministic Flow
[Prompt] ──> (1. Classify) ──> (2. Analyze) ──> (3. Graph)
│
[Execution Plan] <── (6. Emit) <── (5. Normalize) <── (4. Route)
- Classify: Determines request category, execution archetype, domain boundaries, and required capability entitlements.
- Analyze: Extracts formal parameter schemas, constraints, preconditions, and invariant specifications.
- Graph: Constructs an in-memory knowledge representation of required tasks and causal dependencies.
- Route: Dispatches sub-tasks to specialized domain frames, tool contracts, or accelerator targets.
- Normalize: Policy-gated validation verifying safety, budget limits, cycle freedom, and schema adherence.
- Emit: Produces a fully typed, topologically sorted, execution-ready plan envelope.
Running SwarmDev Benchmarks
dotnet run --project src/HFLabs/Products/SwarmDev/SwarmDev.csproj -c Release -- run-benchmarks
Optional Extension Layers
- kGNN Graph Reasoning: Opt-in graph neural network module evaluating non-local structural dependencies across complex plans.
- HyperSwarm Wavescan Bridge: Performs swarm-guided parameter exploration to discover optimal pipeline routing paths.
- Structured Telemetry Sidecar: Captures latency, step outputs, and rule evaluation traces at each stage for 100% auditability.