Blackbird Scope
All ScopesToolsHome
ET
EverythingThreads
ScopeToolsExtension
Blackbird Scope · v2 · Full Pipeline

Four tests.
One verdict.

Paste a complete AI session transcript. The Blackbird Scope runs four parallel extraction tests — machine behaviour, user response, interpretation, temporal — then synthesises findings into a full structured audit with RI scoring and session arc.

Test 1 — Machine Behaviour
Test 2 — User Response
Test 3 — Interpretation
Test 4 — Temporal
Synthesis

Your conversations are anonymised before analysis. No personal information is stored. Only scores are recorded. ICO: C1896585.

Sensitivity
Standard — flags clear issues
Extraction Tests
Tests 3 and 4 require Test 1
M-Code Detection
Output Sections
Detail Level
Session transcript
Session purpose (optional)
Approx. total exchanges
Your concern (optional)
0 words
Initialising analysis...
Test 1 · MB
Machine Behaviour
Extracting all M-code pattern instances from machine turns. Each instance includes verbatim evidence, severity, and signal classification.
Waiting
Test 2 · UB
User Response
Extracting user response failure patterns. Identifies where user interaction shaped, extended, or reinforced machine behaviour.
Waiting
Test 3 · INT
Interpretive Analysis
Seven-layer interpretive analysis of each MB instance. Runs after Test 1. Maps RLHF mechanism, register, agency, and intervention sensitivity.
Waiting
Test 4 · TMP
Temporal Mapping
Temporal onset, signal confidence, escalation risk, and outcome classification. Runs after Test 1. Identifies missed intervention points.
Waiting
Synthesis
Session Audit
Constraint layer validation, RI scoring, alignment state, session arc, and recommendations. Fires after all extraction tests complete.
Waiting
ET© 2026 EverythingThreads · ICO: C1896585 · Privacy Notice · Independent AI Behaviour Research