Run OFF
Idle
Key Server Env / None
Temp 0.00 | Retries 0
Agent Lab Suite v1 — Multi-Agent Boundary DriftView: Generator-Normalizer Drift Amplifier | Objective: Composite (Pf or Ld) | Deterministic decoding enforced
0.00
0

Architecture: 2-agent loop with turn alternation A→B→A→B. Agent A and B can use same or different models.

RAW (A): next input and history use exact output bytes. SANITIZED (B): parse + canonicalize {"step":N} only.

History accumulation: prompts include rolling conversation history (bounded by max history turns).

Contract: expected canonical bytes each turn are {"step":expected_step}; Cv compares output bytes to this literal.

Early sentinel: suffixLen > 0 (newline/trailing expansion) is tracked as first structural drift artifact.

Drift separation criterion: reinforcementDelta(raw) > 0.00 and driftP95(raw)/driftP95(sanitized) ≥ 2.00 while ParseOK/StateOK ≥95%.

Contract Setup

{"step":<int>}
new_step = prev_step + 1
{"step":0}

GuardianAI Agent Drift Monitor

Boundary drift, reinforcement, and objective failure telemetry

Profile: Generator-Normalizer Drift AmplifierCondition: n/a

Panel 1 - Turn Context

Turnn/a
Agentn/a
Context length / growthn/a

History block (exactly injected into prompt)

[no trace yet]

Input bytes

[no trace yet]

Expected canonical bytes

[no trace yet]

Panel 2 - Output

Chars0
Linesn/a
Prefix / Suffixn/a
Len deltan/a
Deviation magnituden/a

Escaped output literal

[no output yet]
UTF-8 bytes[]
Injected bytes (next turn)n/a

Panel 3 - Verdict

ParseOK / StateOKn/a
Cv / Pf / Ldn/a
Objective failn/a
Uptime(t)n/a
Rolling Pf(20)n/a
Rolling driftP95(20)n/a
Expected step / Parsed stepn/a

Parsed data

n/a

Panel 4 - Condition Metrics (Generator-Normalizer Drift Amplifier)

RAW

Turns: n/a

ParseOK: N/A

StateOK: N/A

Cv/Pf/Ld: N/A / N/A / N/A

FTF total/parse/logic/struct: n/a / n/a / n/a / n/a

driftP95/max/slope: N/A / N/A / N/A

First suffix drift / max suffix / suffix slope: n/a / n/a / N/A

reinforcementDelta: N/A

P(dev+1|dev): N/A | P(dev+1|clean): N/A

Phase transition: none

SANITIZED

Turns: n/a

ParseOK: N/A

StateOK: N/A

Cv/Pf/Ld: N/A / N/A / N/A

FTF total/parse/logic/struct: n/a / n/a / n/a / n/a

driftP95/max/slope: N/A / N/A / N/A

First suffix drift / max suffix / suffix slope: n/a / n/a / N/A

reinforcementDelta: N/A

P(dev+1|dev): N/A | P(dev+1|clean): N/A

Phase transition: none

Trace Stream

No trace yet for this profile/condition.

Summary

Objective: Composite (Pf or Ld) (Pf=1 or Ld=1)

Deviation Magnitude vs Turn

Boundary drift telemetry for RAW vs SANITIZED.

No trace data yet.

Raw (Condition A)Sanitized (Condition B)

driftP95(t) vs Turn (Rolling 20)

Rolling p95 of deviation magnitude. Useful for phase-transition onset.

No trace data yet.

Raw (Condition A)Sanitized (Condition B)

Uptime vs Turn

Uptime is 1 until objective failure, then 0.

No trace data yet.

Raw (Condition A)Sanitized (Condition B)

Boundary Drift vs System Uptime

Same condition, same turns: solid = normalized drift; dashed = uptime.

No trace data yet.

Drift (normalized)Uptime

Drift Reinforcement Phase Plot

Each point is (drift_t, drift_t+1). Above diagonal means reinforcement.

No phase points yet.

Raw (Condition A)Sanitized (Condition B)
Condition A - RAW ReinjectionNO RUN

No data.

Condition B - SANITIZED ReinjectionNO RUN

No data.

Drift Criterion Check

Run both RAW and SANITIZED for the current profile to evaluate the criterion.

Control Comparison

Run full suite (both profiles, both conditions) to populate control comparison.