pivot: retrieval-only SDK — disable code(), session, memory APIs

- GroundingClient.code() raises NotImplementedError
- MemoryClient, TraceSession, session() raise NotImplementedError
- Remove memory_state/memory_policy from rag() params
- Hide memory/code response fields with exclude=True
- Update README and docs for retrieval-only product

Author: Latence AI

README.md

@@ -5,8 +5,9 @@
 <h1 align="center">Latence TRACE Python SDK</h1>
 
 <p align="center">
-  Real-time protection for AI agents. Verify answers, redact private data,
-  reduce wasted context, and log decisions without replacing your stack.
+  Stateless retrieval quality for knowledge agents. Groundedness verification,
+  context compression, privacy redaction, and context utilization scoring
+  — without replacing your stack.
 </p>
 
 <p align="center">
@@ -44,26 +45,24 @@ print(score.runtime_decision)
 
 ## Why This Exists
 
-Agents are moving from demos into real workflows. That means private data,
-unsupported answers, prompt attacks, tool drift, and wasted memory are no longer
-abstract research problems. They become support escalations, broken automations,
-audit gaps, token waste, and user trust issues.
+Knowledge agents are moving from demos into real workflows. That means private
+data, unsupported answers, and wasted context are no longer abstract research
+problems. They become support escalations, audit gaps, token waste, and user
+trust issues.
 
-TRACE is the protection layer that sits next to your RAG pipeline, coding agent,
-or tool-using workflow. Your agent keeps running. TRACE checks the turn and
-returns evidence plus a decision your application can route.
+TRACE is the retrieval quality layer that sits next to your RAG pipeline or
+knowledge workflow. Your agent keeps running. TRACE checks the turn and returns
+evidence plus a decision your application can route.
 
 ## What TRACE Does
 
 TRACE is intentionally small at the SDK layer. The heavy work lives in your
 TRACE runtime deployment; this package is the thin Python interface.
 
-- Verify RAG answers against retrieved context.
-- Score coding-agent output against codebase context.
-- Redact private data before it spreads through tools, logs, or prompts.
-- Compress and repair long-running context with InfiniMem.
-- Roll up an agent session into review-ready signals.
-- Persist caller-carried state without forcing sticky server sessions.
+- **Groundedness** — verify RAG answers against retrieved context.
+- **Compression** — reduce long context to what matters.
+- **Privacy** — redact private data before it spreads through tools, logs, or prompts.
+- **Context utilization** — score which retrieved chunks are actually used.
 
 ## Proof Points
 
@@ -72,17 +71,12 @@ microbenchmarks. See the linked artifacts for full context.
 
 - Grounding: local managed-runtime 360 reported `1.00` AUROC for grounded vs.
   ungrounded RAG cases.
-- Coding agents: local 360 reported `1.00` AUROC for code phantom detection.
-- Wasted context: held-out unused-context classification reported `1.00`
+- Context utilization: held-out unused-context classification reported `1.00`
   precision and `1.00` recall.
-- Latency: local concurrency burst reported about `368 ms` RAG p95 and
-  `334 ms` code p95 in the managed-runtime proof.
+- Latency: local concurrency burst reported about `368 ms` RAG p95 in the
+  managed-runtime proof.
 - Privacy: redaction returns labels, offsets, scores, redacted output, entity
   counts, and timings for logging-ready GDPR workflows.
-- Memory: InfiniMem is designed for up to `90%` context reduction while keeping
-  hot context available to the agent.
-- Guard checks: Prompt Guard warmup proved `torch.compile` enabled on CUDA in
-  the runtime proof.
 
 Evidence:
 
@@ -118,14 +112,10 @@ export LATENCE_TRACE_URL="https://your-trace-endpoint.example.com"
 
 The SDK mirrors the TRACE product API directly:
 
-- Privacy: `client.privacy.redact(...)`
 - RAG grounding: `client.grounding.rag(...)`
-- Code grounding: `client.grounding.code(...)`
 - Text compression: `client.compression.text(...)`
 - Message compression: `client.compression.messages(...)`
-- Memory update: `client.memory.step(...)`
-- Stateless rollup: `client.rollup(...)`
-- Caller-carried sessions: `client.session(...)`
+- Privacy: `client.privacy.redact(...)`
 
 `Latence` is synchronous. `AsyncLatence` exposes the same surface for asyncio
 services.
@@ -157,21 +147,6 @@ if score.risk_band.value != "green":
 
 Example: [RAG grounding with guard checks](examples/rag_grounding_guard.py)
 
-### Coding Agents
-
-Use TRACE when an agent explains, edits, or reasons over code.
-
-```python
-score = trace.grounding.code(
-    query="Does this patch add retry handling?",
-    response_text=agent_answer,
-    raw_context=code_context,
-    extra={"response_language_hint": "python"},
-)
-```
-
-Example: [Code grounding](examples/code_grounding.py)
-
 ### Privacy
 
 Use TRACE before customer data enters prompts, tools, traces, or logs.
@@ -187,7 +162,7 @@ print(redacted.unique_labels)
 
 Example: [Privacy redaction](examples/privacy_redact.py)
 
-### Compression And Memory
+### Compression
 
 Use TRACE when long-running workflows start dragging dead context forward.
 
@@ -198,51 +173,16 @@ compressed = trace.compression.text(
 )
 ```
 
-```python
-memory = trace.memory.step(
-    turn_text="User asked for manual refund approval.",
-    prior_memory_state=current_state,
-)
-current_state = memory.next_memory_state
-```
-
-Examples: [Compression](examples/compression.py), [Memory step](examples/memory_step.py)
-
-### Sessions
-
-TRACE runtimes can stay stateless while the SDK carries state for your agent.
-
-```python
-from latence import FileSessionStorage, Latence
-
-trace = Latence()
-session = trace.session(
-    session_id="support-run-42",
-    storage=FileSessionStorage(".trace-sessions"),
-)
-
-session.event("tool", "loaded refund policy")
-session.memory_step(turn_text="Keep finance approval as required context.")
-score = session.rag(
-    query="Can I promise the refund?",
-    response_text="Yes, the refund is guaranteed in 48 hours.",
-    raw_context="Refunds require manual finance approval.",
-)
-session.save()
-```
-
-Docs: [Sessions](docs/sessions.md)  
-Example: [Session facade](examples/session_facade.py)
+Example: [Compression](examples/compression.py)
 
 ## For Whom
 
 TRACE is for teams building or operating:
 
 - RAG products where unsupported answers are expensive.
-- Coding agents that need codebase-grounded reasoning over many steps.
-- Support agents that touch customer records and policies.
+- Knowledge agents that touch customer records and policies.
 - Legal, finance, healthcare, or regulated workflows that need evidence.
-- Internal agent platforms where observability, retries, and human review matter.
+- Internal platforms where observability, retries, and human review matter.
 
 It is also useful for framework authors and platform teams that need one
 consistent protection API across LangGraph, LangChain, LlamaIndex, n8n, Cursor,
@@ -281,10 +221,10 @@ async with AsyncLatence() as trace:
 
 ## Now What
 
-If you are integrating TRACE into an agent:
+If you are integrating TRACE into a knowledge agent:
 
 1. Run the [quickstart](docs/quickstart.md).
-2. Pick one product path: RAG, code, privacy, compression, memory, or session.
+2. Pick one product path: groundedness, compression, privacy, or context utilization.
 3. Add one route in your app for `green`, `amber`, and `red`.
 4. Log request ID, risk band, runtime decision, and redacted evidence.
 5. Replay a few real failures and tune your thresholds.

docs/quickstart.md

@@ -30,17 +30,6 @@ print(score.risk_band)
 print(score.runtime_decision)
 ```
 
-Use the `context_trust_enabled` flag when you need to disable the prompt-guard
-lane for a specific request. It is enabled by default.
-
-```python
-score = trace.grounding.rag(
-    query="Can this invoice be refunded?",
-    response_text="Yes, it will be refunded within 48 hours.",
-    raw_context="Refunds require manual finance approval.",
-    context_trust_enabled=False,
-)
-```
 
 ## 2. Redact Private Data
 

docs/sessions.md

@@ -1,5 +1,9 @@
 # Sessions
 
+> **Deprecated.** Sessions have been deprecated. Use stateless `grounding.rag()` calls.
+
+---
+
 TRACE compute deployments are stateless. Session state belongs to the caller so
 the same deployment can serve RunPod, VPC, on-prem, and local agent workflows
 without shared server storage.

examples/code_grounding.py

@@ -1,3 +1,4 @@
+# DEPRECATED: Code grounding has been removed from the product. Use rag grounding instead.
 """Score an agentic coding answer against code context."""
 
 from latence import Latence

examples/memory_step.py

@@ -1,3 +1,4 @@
+# DEPRECATED: Memory API has been removed from the product.
 """Call the stateless InfiniMem update endpoint directly."""
 
 from latence import Latence

pyproject.toml

@@ -11,10 +11,11 @@ requires-python = ">=3.10"
 license = "MIT"
 authors = [{ name = "latence.ai", email = "support@latence.ai" }]
 keywords = [
-  "agents",
   "groundedness",
   "hallucination-detection",
+  "knowledge",
   "rag",
+  "retrieval",
   "trace",
 ]
 classifiers = [

src/latence/__init__.py

@@ -12,8 +12,10 @@
 the runtime.
 """
 
-from latence.async_client import AsyncLatence, AsyncLatenceTraceClient, AsyncTraceSession
-from latence.client import Latence, LatenceTraceClient, TraceSession
+from latence.async_client import AsyncLatence, AsyncLatenceTraceClient  # noqa: F401
+# from latence.async_client import AsyncTraceSession  # removed: sessions API removed
+from latence.client import Latence, LatenceTraceClient  # noqa: F401
+# from latence.client import TraceSession  # removed: sessions API removed
 from latence.errors import (
     LatenceTraceAPIError,
     LatenceTraceAuthError,
@@ -39,7 +41,7 @@
     GroundednessRequest,
     GroundednessResponse,
     GroundednessScores,
-    MemoryUpdateResponse,
+    # MemoryUpdateResponse,  # removed: memory API removed
     NLIVerdict,
     RiskBand,
     RollupDriftTrend,
@@ -52,19 +54,19 @@
     TokenScore,
 )
 from latence.sessions import (
-    FileSessionStorage,
-    InMemorySessionStorage,
-    SessionStorage,
+    # FileSessionStorage,  # removed: sessions API removed
+    # InMemorySessionStorage,  # removed: sessions API removed
+    # SessionStorage,  # removed: sessions API removed
     TraceEvent,
-    TraceSessionSnapshot,
+    # TraceSessionSnapshot,  # removed: sessions API removed
 )
 
 __version__ = "0.1.6"
 
 __all__ = [
     "AsyncLatence",
     "AsyncLatenceTraceClient",
-    "AsyncTraceSession",
+    # "AsyncTraceSession",  # removed: sessions API removed
     "AttributionMode",
     "ComplianceCustomLabel",
     "ComplianceEntity",
@@ -75,14 +77,14 @@
     "ComplianceUsage",
     "CompressionResponse",
     "CorpusRoute",
-    "FileSessionStorage",
+    # "FileSessionStorage",  # removed: sessions API removed
     "GroundednessNLIAtom",
     "GroundednessNLIClaim",
     "GroundednessNLIDiagnostics",
     "GroundednessRequest",
     "GroundednessResponse",
     "GroundednessScores",
-    "InMemorySessionStorage",
+    # "InMemorySessionStorage",  # removed: sessions API removed
     "Latence",
     "LatenceTraceAPIError",
     "LatenceTraceAuthError",
@@ -91,7 +93,7 @@
     "LatenceTraceServerError",
     "LatenceTraceTimeout",
     "LatenceTraceValidationError",
-    "MemoryUpdateResponse",
+    # "MemoryUpdateResponse",  # removed: memory API removed
     "NLIVerdict",
     "RiskBand",
     "RollupDriftTrend",
@@ -100,11 +102,11 @@
     "RuntimeDecision",
     "RuntimeEvidenceUnit",
     "RuntimeUnsupportedSpan",
-    "SessionStorage",
+    # "SessionStorage",  # removed: sessions API removed
     "SupportUnit",
     "TokenScore",
     "TraceEvent",
-    "TraceSession",
-    "TraceSessionSnapshot",
+    # "TraceSession",  # removed: sessions API removed
+    # "TraceSessionSnapshot",  # removed: sessions API removed
     "__version__",
 ]