agent-learning 0.6.0__tar.gz → 0.7.0__tar.gz

{agent_learning-0.6.0/src/agent_learning.egg-info → agent_learning-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-learning
-Version: 0.6.0
+Version: 0.7.0
 Summary: Evidence-driven decision learning for AI agents with inspectable autonomy gates.
 Author: Chris Tava
 License: MIT License
@@ -87,11 +87,15 @@ The SDK improves agents without LLM weight fine-tuning. There are no GPU fine-tu
 action plus historical correctness, reward, result summaries, and per-metric
 quality feedback for the agent to use on its next delegated decision.
 
-It also returns an evidence-gated autonomy assessment. Routine confirmation
-stops only after the recommended action has enough scored outcomes, a 95%
-Wilson correctness lower bound of at least 90%, positive mean reward, strong
-probability and margin, and the same winner across three trained snapshots.
-Autonomous executions continue learning from observable outcomes, while a
-configurable sample requests user feedback to detect drift.
+It also returns a complexity-proportional autonomy assessment. A persisted
+profile covers intent ambiguity, context variability, outcome observability,
+decision impact, reversibility, and mandatory approval; action-space size is
+derived. The resulting low, standard, high, or critical tier scales required
+outcomes, Wilson confidence, reward, probability, margin, stable snapshots, and
+drift-audit rate. Autonomous executions continue learning from observable
+outcomes, while tier-scaled samples request user feedback to detect drift.
+An explicit accepted-feedback episode is a separate durable authorization path:
+it pins that action for the task policy and suppresses future feedback prompts
+until the user explicitly rejects it.
 
 Every episode, reward, run, and deployment is captured by the configured store — in-memory or local files by default, or Azure Cosmos DB — giving you a complete lineage and audit trail of how the policy evolved over time.

{agent_learning-0.6.0 → agent_learning-0.7.0}/PYPI.md

@@ -20,11 +20,15 @@ The SDK improves agents without LLM weight fine-tuning. There are no GPU fine-tu
 action plus historical correctness, reward, result summaries, and per-metric
 quality feedback for the agent to use on its next delegated decision.
 
-It also returns an evidence-gated autonomy assessment. Routine confirmation
-stops only after the recommended action has enough scored outcomes, a 95%
-Wilson correctness lower bound of at least 90%, positive mean reward, strong
-probability and margin, and the same winner across three trained snapshots.
-Autonomous executions continue learning from observable outcomes, while a
-configurable sample requests user feedback to detect drift.
+It also returns a complexity-proportional autonomy assessment. A persisted
+profile covers intent ambiguity, context variability, outcome observability,
+decision impact, reversibility, and mandatory approval; action-space size is
+derived. The resulting low, standard, high, or critical tier scales required
+outcomes, Wilson confidence, reward, probability, margin, stable snapshots, and
+drift-audit rate. Autonomous executions continue learning from observable
+outcomes, while tier-scaled samples request user feedback to detect drift.
+An explicit accepted-feedback episode is a separate durable authorization path:
+it pins that action for the task policy and suppresses future feedback prompts
+until the user explicitly rejects it.
 
 Every episode, reward, run, and deployment is captured by the configured store — in-memory or local files by default, or Azure Cosmos DB — giving you a complete lineage and audit trail of how the policy evolved over time.

{agent_learning-0.6.0 → agent_learning-0.7.0}/README.md

@@ -13,7 +13,7 @@ user-visible outcome, scores the evidence, and applies a small CPU update to
 the next decision.
 
 <p align="center">
-   <img src="images/agent-decision-making.svg" alt="Animated agentic decision loop: choose a bounded action, execute it, score the observed outcome, and improve the next decision" width="960" style="max-width:100%; height:auto;" />
+   <img src="images/agent-decision-making.svg" alt="Animated agentic decision loop: choose, execute, score, improve, then become autonomous through explicit user acceptance or proportional statistical evidence" width="960" style="max-width:100%; height:auto;" />
 </p>
 
 ## The decision loop
@@ -22,9 +22,10 @@ the next decision.
    least two actions the agent can actually execute. A TaskPolicy owns the
    probability distribution for that `(agent_id, task_id)` pair.
 
-2. **Choose and execute.** `task-policy-decide` samples an action from the
-   active softmax policy and returns learned feedback from earlier attempts.
-   The agent uses that feedback and executes the selected action.
+2. **Choose and execute.** `task-policy-decide` samples from the active softmax
+   policy while supervised, then uses the stable winner greedily after autonomy
+   is earned. The agent consumes learned feedback and executes the selected
+   action.
 
 3. **Observe and score.** A completed episode preserves the decision context,
    selected action, output, result summary, latency, and independently
@@ -36,12 +37,12 @@ the next decision.
    probability, worse-than-usual choices lose probability, and exploration
    remains available.
 
-5. **Earn autonomy.** `task-policy-decide` stops routine confirmation only after
-   the recommended action passes every evidence gate: scored outcomes, Wilson
-   correctness confidence, positive reward, action probability, probability
-   margin, and three stable trained snapshots. Autonomous actions still record
-   observable outcomes, and a configurable sample requests user feedback to
-   detect drift.
+5. **Earn or grant autonomy.** `task-policy-decide` stops routine confirmation
+   when the recommended action passes every evidence gate or the user explicitly
+   accepts an action for that task policy. Explicit acceptance is durable, pins
+   the accepted action, and disables repeat feedback prompts until a later
+   rejection. Statistical thresholds and drift audits still scale with the
+   persisted complexity profile. Autonomous actions record observable outcomes.
 
 Everything runs in the existing Python process. Scoring is local by default;
 configured Azure AI evaluators remain available as an opt-in. Stores can be
@@ -71,6 +72,10 @@ never consumed.
 The response also contains an `autonomy` block. Agents must follow
 `execute_without_confirmation`, `request_user_feedback`, and
 `outcome_recording` rather than inventing their own confidence threshold.
+`autonomy.complexity` explains the declared profile, derived action-space
+points, risk floors, tier, and resolved proportional criteria.
+`authorization_basis` distinguishes explicit `user_acceptance` from
+`statistical_evidence`.
 
 ## Install
 
@@ -121,6 +126,20 @@ Define actions the agent can really execute in `actions.json`:
 ]
 ```
 
+Declare intent and decision complexity in `complexity.json`:
+
+```json
+{
+   "intent_ambiguity": "medium",
+   "context_variability": "variable",
+   "outcome_observability": "direct",
+   "decision_impact": "medium",
+   "reversibility": "costly",
+   "requires_human_approval": false,
+   "rationale": "Repository search is bounded and directly observable."
+}
+```
+
 Initialize the decision once, then ask the active policy what to execute:
 
 ```powershell
@@ -128,7 +147,8 @@ agent-learn task-policy-init `
    --agent-id code-reviewer `
    --task-id choose-repository-search `
    --decision-context "Choose a repository search strategy for a coding task" `
-   --actions .\actions.json
+   --actions .\actions.json `
+   --complexity-profile .\complexity.json
 
 agent-learn task-policy-decide `
    --agent-id code-reviewer `
@@ -165,6 +185,23 @@ python tests/functional_cli_interactive.py
 python tests/functional_cli_batch.py
 ```
 
+## Visualizing autonomy
+
+The animated SVG above shows the full five-stage loop. The retained
+[learning-loop SVG](images/agent-learning-loop.svg) keeps the compact
+TaskPolicy → Score → Learner view and places explicit acceptance and statistical
+evidence routes on its return path.
+
+Render the decision animation with Manim:
+
+```powershell
+.\animiations\run_animation.ps1 `
+   -Script decision_making_animation.py `
+   -Scene AgentDecisionMaking `
+   -Quality h `
+   -Preview
+```
+
 ## Usage
 
 The `agent-learn` CLI provides the decision lifecycle and inspection
@@ -176,7 +213,8 @@ agent-learn --version
 agent-learn tasks-list <agent_id> [--decision-only]
 agent-learn task-episodes-count <agent_id> [--task-id <task_id>] [--include-incomplete] [--start-date <date>] [--end-date <date>]
 agent-learn task-episodes-list <agent_id> [--task-id <task_id>] [--limit <1-500>] [--include-incomplete] [--start-date <date>] [--end-date <date>]
-agent-learn task-policy-init --agent-id <agent_id> --task-id <task_id> --decision-context <context> --actions ./actions.json
+agent-learn task-policy-init --agent-id <agent_id> --task-id <task_id> --decision-context <context> --actions ./actions.json [--complexity-profile ./complexity.json]
+agent-learn task-policy-complexity-set --agent-id <agent_id> --task-id <task_id> --profile ./complexity.json
 agent-learn task-policy-decide --agent-id <agent_id> --task-id <task_id> [--history-limit <1-500>] [--greedy] [--seed <integer>]
 agent-learn task-episode-register --agent-id <agent_id> --task-id <task_id> --episode ./episode.json [--require-decision-policy]
 agent-learn score --agent-id <agent_id> [--task-id <task_id>] [--limit <1-500>]
@@ -188,6 +226,8 @@ agent-learn task-policy --agent-id <agent_id> --task-id <task_id>
 
 - [Agentic decision making](docs/decision-making.md): concepts, evidence,
    workflow, math, and deployment.
+- [Complexity-proportional autonomy](docs/autonomy-complexity.md): declared
+   complexity dimensions, tier calculation, risk floors, and scaled gates.
 - [Scout decision integration](docs/scout-agent-learn-skill.md): apply learned
    delegated decisions during execution.
 - [Scout decision training](docs/scout-automation-agent-learning.md): train

{agent_learning-0.6.0 → agent_learning-0.7.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "agent-learning"
-version = "0.6.0"
+version = "0.7.0"
 description = "Evidence-driven decision learning for AI agents with inspectable autonomy gates."
 readme = "PYPI.md"
 requires-python = ">=3.10"

{agent_learning-0.6.0 → agent_learning-0.7.0}/src/agent_learning/__init__.py

@@ -56,7 +56,14 @@ Quick start::
 """
 
 from ._version import __version__
-from .autonomy import AutonomyAssessment, assess_autonomy, wilson_lower_bound
+from .autonomy import (
+    AutonomyAssessment,
+    ComplexityAssessment,
+    ComplexityProfile,
+    assess_autonomy,
+    assess_complexity,
+    wilson_lower_bound,
+)
 from .capture import CaptureContext, EpisodeCapture, get_capture
 from .classifiers import (
     AdherenceScorer,
@@ -68,6 +75,7 @@ from .classifiers import (
 )
 from .config import (
     AutonomyConfig,
+    AutonomyTier,
     CaptureConfig,
     CosmosConfig,
     ScoreConfig,
@@ -113,6 +121,7 @@ __all__ = [
     "Action",
         "AutonomyAssessment",
         "AutonomyConfig",
+        "AutonomyTier",
     "AdherenceScorer",
     "AgentSummary",
     "AgentTaskSummary",
@@ -121,6 +130,8 @@ __all__ = [
     "Classifier",
     "ClassifierResult",
     "CompletionScorer",
+    "ComplexityAssessment",
+    "ComplexityProfile",
     "ContextualSoftmaxPolicy",
     "CosmosConfig",
     "CosmosStore",
@@ -157,6 +168,7 @@ __all__ = [
     "TrainingStatus",
     "__version__",
     "assess_autonomy",
+    "assess_complexity",
     "default_metrics",
     "evaluate_all",
     "get_capture",

{agent_learning-0.6.0 → agent_learning-0.7.0}/src/agent_learning/_version.py

@@ -1,3 +1,3 @@
 """Package version."""
 
-__version__ = "0.6.0"
+__version__ = "0.7.0"