Update README, version, changelog, dependency bounds, and limitations

CHANGELOG.md

@@ -1,7 +1,11 @@
 # Changelog
 
-## Unreleased
+## 0.1.0 (2026-04-19)
 
+Public launch release.
+
+- README: lead with clone + uv install, BSL 1.1 framing, visible limitations
+- Tightened `litellm` dependency bounds (`>=1.0,<2.0,!=1.82.7,!=1.82.8`)
 - Evolver protocol: unified Reflector + GEPA + Paradigm behind `Evolver` interface
 - OpenClaw adapter: transparent proxy for any OpenAI-compatible agent
 - n8n integration: webhook-based learning for workflow platforms

README.md

@@ -13,6 +13,10 @@ agent-environment interactions, learns from them, and feeds improvements back
 into the agent. Three learning layers — **harness**, **router**, **weights** —
 all following the same protocol.
 
+> **Source-available under [BSL 1.1](LICENSE).** Free for dev, test, research,
+> and production use below $10 M revenue. Converts to Apache 2.0 on
+> April 1, 2030. See [License](#license) for details.
+
 <p align="center">
   <img src="architecture.png" alt="ClawLoop Architecture" width="720">
 </p>
@@ -205,18 +209,17 @@ and an `EpisodeSummary` containing reward signals. See `clawloop/environments/ma
 
 </details>
 
-<details>
-<summary><strong>Limitations</strong></summary>
+## Limitations
 
+- **Harness/playbook learning is the stable, recommended path.** Router and
+  weight layers work but have more constraints — see below.
 - **`mode="full"`** (simultaneous harness + weight training) is disabled.
   The on-policy boundary after harness updates needs rework for GRPO advantage
   computation. Use `weight` and `harness_learning` separately for now.
 - **Episode construction is manual.** There is no `ProblemEnv` base class yet.
   New environments must build `Episode` objects directly. A higher-level
   abstraction (like Tinker cookbook's `ProblemEnv`) is planned.
 
-</details>
-
 ## Enterprise
 
 ClawLoop Enterprise adds premium learning backends and managed

clawloop/__init__.py

@@ -1,6 +1,6 @@
 """ClawLoop — Learning from Experience unified learning API."""
 
-__version__ = "0.0.1"
+__version__ = "0.1.0"
 
 from clawloop.agent import ClawLoopAgent
 from clawloop.archive import (

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "clawloop"
-version = "0.0.1"
+version = "0.1.0"
 description = "AI agents that learn from experience — observe, learn, improve"
 requires-python = ">=3.12"
 license = {file = "LICENSE"}
@@ -20,7 +20,7 @@ classifiers = [
     "Programming Language :: Python :: 3.13",
 ]
 dependencies = [
-    "litellm>=1.0",
+    "litellm>=1.0,<2.0,!=1.82.7,!=1.82.8",
     "pydantic>=2.0",
     "python-dotenv>=1.0",
 ]