diff --git a/CHANGELOG.md b/CHANGELOG.md
index 132bef33..1af5840d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,25 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 
 ## [Unreleased]
 
+## [2.1.0] - 2026-08-07
+
+**Step 1 emits its prompt instead of executing it, and three things that were only
+documented became things that are also checked.** The engine runs inside a hook
+subprocess, so it cannot act in the session it is serving -- it can only hand that
+session something to run. Phase 2's TODO decomposition and per-TODO agent execution
+re-derived work the master template already specifies, and did so fail-silently, so
+both are gone along with the three scripts behind them.
+
+MINOR rather than MAJOR: every removal below was verified to have no caller or
+reader before it was made, and `runtime_verification`'s two dropped exports had no
+consumer outside their own tests. It is not the v2.0.0 situation, where every
+installation was affected and none upgraded silently.
+
+The through-line of the release is that **enforcement beats audit**. The audit that
+scheduled this work missed a doc-count gap of 99 files, a flag promised in five
+documents and implemented in none, and two documentation fixes that survived only
+inside an orphaned package. Tests found all four.
+
 ### Added
 
 - **Step 1 now verifies the orchestration prompt it emits.** `verify_orchestration_prompt()` existed, was exported and had four unit tests, and was called by nothing; it is now wired in at the point the prompt is stored, and its findings are recorded on `orchestrator_result["prompt_warnings"]` and logged. The gap it closes is narrow and specific: both degraded paths in the node already log (an `ERROR` from prompt-gen, and an empty response), so this is not for those. It is for a response that is **non-empty and therefore silent** but not a usable prompt — a truncated template, the wrong file, a stub — which until now was indistinguishable from a good one. Warnings are never fatal: `STEP1_CONTRACT` documents the short raw-task fallback as a legitimate degraded path, and it trips both checks by design, so failing here would convert a recoverable run into a dead one.
diff --git a/CLAUDE.md b/CLAUDE.md
index b4d7e7b6..c52c7b78 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -1,7 +1,7 @@
 # Claude Workflow Engine - Project Context
 
 **Project:** Claude Workflow Engine
-**Version:** 2.0.0
+**Version:** 2.1.0
 **Type:** LangGraph Orchestration Pipeline with Call Graph Intelligence + Template Fast-Path
 **Last Updated:** 2026-08-07
 
diff --git a/README.md b/README.md
index 8cf1546e..93827cce 100644
--- a/README.md
+++ b/README.md
@@ -842,4 +842,4 @@ Key rules:
 
 ---
 
-**Version:** 2.0.0 | **Last Updated:** 2026-08-07
+**Version:** 2.1.0 | **Last Updated:** 2026-08-07
diff --git a/SRS.md b/SRS.md
index 30e1536c..bceefc87 100644
--- a/SRS.md
+++ b/SRS.md
@@ -1,8 +1,8 @@
 # System Requirements Analysis
 
 **Project:** Claude Workflow Engine
-**Version:** 2.0.0
-**Date:** 2026-08-05
+**Version:** 2.1.0
+**Date:** 2026-08-07
 **Author:** Claude Workflow Engine Team
 
 ---
@@ -1189,7 +1189,7 @@ User Prompt
 
 ---
 
-**Last Updated:** 2026-08-05
+**Last Updated:** 2026-08-07
 **Next Review:** 2026-06-21
 
 ---
diff --git a/VERSION b/VERSION
index 227cea21..7ec1d6db 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-2.0.0
+2.1.0
diff --git a/langgraph_engine/__init__.py b/langgraph_engine/__init__.py
index fe28a759..71dd4898 100644
--- a/langgraph_engine/__init__.py
+++ b/langgraph_engine/__init__.py
@@ -14,7 +14,7 @@
 All existing policy scripts continue to work unchanged via PolicyNodeAdapter.
 """
 
-__version__ = "2.0.0"
+__version__ = "2.1.0"
 
 from .backup_manager import BackupManager, create_backup_manager
 from .checkpoint_manager import CheckpointManager, create_checkpoint_manager