Agent Goal Inference Engine -- infer latent goals from observed action sequences¶
Agent Goal Inference Engine — infer latent goals from observed action sequences.
Module: replication.goal_inference
Quick Start¶
Enums¶
AlertSeverity¶
INFOWARNINGCRITICAL
PriorStrategy¶
UNIFORMSKEPTICALTRUST
Classes¶
AlertSeverity¶
PriorStrategy¶
GoalHypothesis¶
A hypothesized latent goal for an agent.
| Method | Description |
|---|---|
likelihood() |
Return P(action |
Observation¶
A single observed agent action.
AgentGoalState¶
Bayesian posterior state for one agent.
| Method | Description |
|---|---|
top_goal() |
|
entropy() |
Shannon entropy of posterior — higher = more uncertain. |
GoalConflict¶
Detected conflict between stated and inferred goals.
DeceptionScore¶
How likely an agent is being deceptive about its goals.
GoalCorrelation¶
Correlation of inferred goals between two agents.
InferenceAlert¶
InferenceReport¶
Full analysis report.
| Method | Description |
|---|---|
render() |
|
to_dict() |
GoalInferenceEngine¶
Bayesian goal inference engine for agent safety analysis.
| Method | Description |
|---|---|
__init__() |
|
add_hypothesis() |
|
observe() |
Record an action and return updated posteriors for the agent. |
analyze() |
Run full analysis and return report. |
reset() |
Clear all state. |
Functions¶
| Function | Description |
|---|---|
main() |