PASR Local

PASR Local helps evaluate proposed interventions before they become effects. It produces bounded decisions, evidence / readouts, intervention memory, and local continuity material. It is a component and runtime inside FoldEngine's containerized products — the main product line — and a proof / evaluator surface. It is not the whole product story by itself.

The current PASR package supports bounded intervention classes only:

• HOST_UI_EVENT — host / UI events
• SIMULATED_EFFECT — simulated effects
• ADAPTER_PROPOSAL — adapter proposals
• Evidence packs
• Human-readable readouts
• Local / container run path
• Mac, Linux, and Pi / Edge no-hardware profiles

PASR's deployment-profile language is now being generalized across the FoldEngine product family through the Product Container Shell standard. Pi / edge is a deployment target, not a product, and is claimed only where validated.

PASR also fits human-facing systems such as prompts, journey transitions, and operator cues where an intervention should be checked before it reaches the person.

PASR Local is deliberately bounded. To avoid overclaiming, here is what it is not:

• It is not a physical actuator controller — physical actuator execution remains outside the current package and requires a separate future safety profile.
• It is not safety-certified.
• It is not a robot controller, and it does not control hardware.
• It is not a cloud service, and there is no cloud dependency for local evaluation.
• It is not a native app or installer yet, and not a C/C++ ABI release yet.
• There is no hardware execution in the current package; the evaluation uses no privileged Docker, no /dev mount, and no host network mode.
• It is not the private FoldEngine Memory engine, and it does not expose FoldEngine Memory internals.
• It is not a raw memory dump, and not resonance / Core3 / Codex exposure.
• It is not a ROS/ROS2 runtime today, and it does not replace ROS/ROS2.
• It does not do ledger writes today.

Stabilized Loop is a shared runtime primitive inside the product line — not a standalone product. PASR substrate, Riverbed, Import Spine, and the SEAM / packet chain are shared components underneath containerized products. PASR Local is the evaluator/runtime package that gates proposed interventions; it does not replace those products or claim to be the whole hierarchy.

You receive a local evaluation artifact and a guided return path:

• PASR release tarball
• SHA256 checksum and manifest
• external evaluator quickstart
• archive verifier
• Docker config / smoke / readout path
• expected BLOCKED demo result
• evaluator return packet workflow
• optional guided review call

Two evaluations

The gate-only demo proves PASR can stop an unready intervention. The memory-loop evaluation proves PASR can remember bounded intervention decisions, reconstruct the shape of repeated decisions, and show the operator what pattern is forming.

An opt-in Docker memory-loop evaluation is available for technical review:

docker compose -f docker-compose.pasr.yml run --rm pasr-memory-loop-eval

Intentionally not included: no hardware execution, not a physical actuator controller, and no cloud dependency for local evaluation.

PASR Local is no longer only a one-off gate. Each gated decision can feed a bounded local memory loop, so PASR can show the operator what pattern is forming across repeated interventions:

1. proposed intervention
2. → PASR gate / evidence
3. → bounded decision event
4. → append-only local memory
5. → geometry snapshot
6. → intervention_memory readout
7. → human / JSON / JSONL export

What the pattern means

• repeated BLOCK events form a crease
• NOT_READY forms a readiness seam
• WAIT_FOR_PLANE_READY becomes the suggested stitch / operator action

See a memory-loop sample output →

The export is not the memory engine; it is a safe projection of what the memory engine knows.

How PASR works: the architecture →

• AI agent / tool builders
• Operator-console teams
• Simulator / lab workflow teams
• Robotics preflight teams
• Local / offline governance evaluators
• Research teams exploring evidence-gated interventions