Protocol Demo v5.0.0 · Live local-LLM governance
Connecting to LM Studio…
OpenAI-compatible · HMAC-signed · Tamper-evident · CRP-* headers

See CRP v5 govern every inference call

This demo drives a local model in LM Studio through the Context Relay Protocol v5 runtime. Watch the canonical CRP-* header namespace, signed session tokens, window-level HMAC chain, CKF memory, DPE risk scoring, safety budget, and compliance classification appear in real time.

💡 Pick a tab, choose a template, and run an action. A tip here explains what CRP just did and what each panel is showing.

Positioned tool loop v5 · SPEC-049/050

CRP v5 turns a small local model into a capable agent by positioning, not injection: the model is placed on one operation at a time and shown only the 1–3 tools that operation needs — it never sees the full catalogue. Each tool result becomes a typed state fact, so the working window stays bounded no matter how many calls run.

Try a template:
Run the loop — only the 1–3 tools each operation needs are surfaced to the model.
Catalogue size
Frame tokens
Observations
Completion
Halted
Profile
Run the loop to see the operation plan.
No events yet.
No observations yet.

Governed dispatch

Send a prompt. CRP returns a governed response plus every governance signal.

Try a template:
Risk
Quality tier
Grounding
Fabrications
Saturation
Safety budget
Tokens

Context Knowledge Fabric

Facts extracted from model answers are stored, recalled, and cited across windows.

0
Total facts
0
This window
0
Recalled
No facts yet. Start a conversation in the Dispatch tab.

Continuation DAG

CRP windows form a DAG: LINEAR, FAN_OUT, FAN_IN, and BRANCH. Each window carries a continuation ID and a window HMAC.

No windows yet.

Tamper-evident audit chain

Every window has a HMAC chained from its parent. Tamper with one window and the chain breaks.

No events yet.

CRP-* header namespace

Every governed response carries the canonical CRP v5 header set.

Policy enforcement

Declare a safety policy and watch CRP halt generation when risk crosses the threshold.

Compliance classification

CRP classifies each response against EU AI Act, GDPR PII, NIST, and ISO 42001 controls.

Multi-agent safety

CRP headers travel with A2A/MCP agent messages. The safety budget acts as a circuit breaker across hops.

Safety budget
Circuit state
Budget health
Loop depth
Tool calls
Windows

Active AI safety

CRP runs Layer 1 validation, Layer 2 injection/PII scanning, DPE risk scoring, policy enforcement, safety budget, and HMAC audit chains. These checks are active: they happen on every request, are surfaced in CRP-* headers, and are recorded in the tamper-evident audit log.

No dispatch yet.
No dispatch yet.

Unlike black-box guardrails, CRP publishes its full safety surface: capabilities, defaults, coverage rules, and explicit out-of-scope items.

Click the button to load the surface map.