chore(release): prepare core 0.14.0 (#1227)

Author: jayscambler

CHANGELOG.md

@@ -4,6 +4,8 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased]
 
+## [0.14.0] - 2026-07-21
+
 ### Added
 
 - TypeScript interactive runs now advertise `agent_task_plan_v1` to transcript clients and emit strict, redacted `task_plan_updated` snapshots as work moves through meaningful execution phases. Plans carry stable run/plan identity, monotonic versions and revisions, sticky completion, explicit replan receipts, terminal step state before the run terminal event, and exact durable replay across reconnects and server restarts. Python parity is deferred until its interactive server can supply the same durable transcript identity and replay guarantees (AC-896).
@@ -676,7 +678,9 @@ A new cross-runtime parity audit (`test_cli_contract_parity.py` + `cli-contract-
 - FastAPI dashboard with WebSocket events.
 - CLI via Typer (Python) and `parseArgs` (TypeScript).
 
-[Unreleased]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.12.0...HEAD
+[Unreleased]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.14.0...HEAD
+[0.14.0]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.13.0...py-v0.14.0
+[0.13.0]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.12.0...py-v0.13.0
 [0.12.0]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.11.0...py-v0.12.0
 [0.11.0]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.10.0...py-v0.11.0
 [0.10.0]: https://github.com/greyhaven-ai/autocontext/compare/py-v0.9.0...py-v0.10.0

README.md

@@ -27,9 +27,9 @@ autocontext is a harness for agent improvement. Give it a goal, it runs the task
 
 | Surface             | Command                               |
 | ------------------- | ------------------------------------- |
-| Python CLI          | `uv tool install autocontext==0.13.0` |
-| Python library/dev  | `uv pip install autocontext==0.13.0`  |
-| TypeScript/Node CLI | `bun add -g autoctx@0.13.0`           |
+| Python CLI          | `uv tool install autocontext==0.14.0` |
+| Python library/dev  | `uv pip install autocontext==0.14.0`  |
+| TypeScript/Node CLI | `bun add -g autoctx@0.14.0`           |
 | Pi extension        | `pi install npm:pi-autocontext@0.9.0` |
 
 The PyPI package is `autocontext`; the CLI is `autoctx`. The npm package is `autoctx` (not the unrelated `autocontext` npm package). Provider variables live in [`.env.example`](.env.example).
@@ -86,10 +86,10 @@ Everything is filesystem-first: inspect it, diff it, replay it, export it, or fe
 Python owns the full control-plane package; TypeScript owns several operator-facing surfaces, the TUI, and Node runtime adapters. Start with [autocontext/README.md](autocontext/README.md) or [ts/README.md](ts/README.md).
 
 <!-- autocontext-whats-new:start -->
-## What's New in 0.13.0
+## What's New in 0.14.0
 
-- **Python safe active-run stopping** brings `safe_run_stop_v1` to the Python interactive server: stop a running or paused run at a cooperative generation boundary, preserve completed generations and best score, and receive one correlated terminal receipt (first terminal outcome wins).
-- **Leaner TypeScript install:** the OpenAI and Anthropic SDKs are now optional peer dependencies, so installing `autoctx` no longer pulls them in transitively. Install the provider SDK you use, for example `npm install autoctx @anthropic-ai/sdk`.
+- **Live TypeScript task plans:** interactive runs now advertise `agent_task_plan_v1` and emit semantic `task_plan_updated` snapshots for initial planning, meaningful progress, replanning, completion, failure, and safe stop.
+- **Durable, privacy-safe replay:** task-plan snapshots use stable identity and monotonic revisions, redact credential-shaped values, reject malformed or oversized frames atomically, and restore exactly across reconnects and server restarts. Python producer parity remains explicitly deferred.
 <!-- autocontext-whats-new:end -->
 
 ## Scenario Families

autocontext/assets/whats_new.txt

@@ -1,2 +1,2 @@
-**Python safe active-run stopping** brings `safe_run_stop_v1` to the Python interactive server: stop a running or paused run at a cooperative generation boundary, preserve completed generations and best score, and receive one correlated terminal receipt (first terminal outcome wins).
-**Leaner TypeScript install:** the OpenAI and Anthropic SDKs are now optional peer dependencies, so installing `autoctx` no longer pulls them in transitively. Install the provider SDK you use, for example `npm install autoctx @anthropic-ai/sdk`.
+**Live TypeScript task plans:** interactive runs now advertise `agent_task_plan_v1` and emit semantic `task_plan_updated` snapshots for initial planning, meaningful progress, replanning, completion, failure, and safe stop.
+**Durable, privacy-safe replay:** task-plan snapshots use stable identity and monotonic revisions, redact credential-shaped values, reject malformed or oversized frames atomically, and restore exactly across reconnects and server restarts. Python producer parity remains explicitly deferred.

autocontext/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "autocontext"
-version = "0.13.0"
+version = "0.14.0"
 description = "autocontext control plane for iterative strategy evolution."
 readme = "README.md"
 license = { text = "Apache-2.0" }

autocontext/src/autocontext/__init__.py

@@ -8,7 +8,7 @@
 
 __all__ = ["AutoContext", "ExtensionAPI", "HookBus", "HookEvents", "HookResult", "__version__"]
 
-__version__ = "0.13.0"
+__version__ = "0.14.0"
 
 
 def __getattr__(name: str) -> Any:

autocontext/src/autocontext/banner.py

@@ -22,7 +22,7 @@
 SYNC_BLOCK_END = "<!-- autocontext-readme-hero:end -->"
 WHATS_NEW_BLOCK_START = "<!-- autocontext-whats-new:start -->"
 WHATS_NEW_BLOCK_END = "<!-- autocontext-whats-new:end -->"
-README_WHATS_NEW_HEADING = "What's New in 0.13.0"
+README_WHATS_NEW_HEADING = "What's New in 0.14.0"
 FALLBACK_BANNER_ART = "autocontext"
 README_WORDMARK_DARK_PATH = "autocontext/assets/autocontext-wordmark-dark.svg"
 README_WORDMARK_LIGHT_PATH = "autocontext/assets/autocontext-wordmark.svg"

autocontext/tests/test_release_surface_manifest.py

@@ -42,7 +42,7 @@ def test_release_manifest_renders_whats_new_asset() -> None:
 def test_pi_version_syncs_install_snippets() -> None:
     sync = _load_sync_module()
     manifest = sync.ReleaseManifest(
-        core_version="0.13.0",
+        core_version="0.14.0",
         pi_version="0.9.0",
         pi_autoctx_dependency="^0.9.0",
         whats_new=("**One** thing.",),
@@ -69,6 +69,6 @@ def test_release_manifest_checks_package_version_files() -> None:
 
     issues = sync.check_release_surfaces(manifest)
 
-    assert "autocontext/src/autocontext/__init__.py version 0.13.0 != manifest 9.9.9" in issues
+    assert "autocontext/src/autocontext/__init__.py version 0.14.0 != manifest 9.9.9" in issues
     assert "pi/package.json version 0.9.0 != manifest 0.8.0" in issues
     assert "pi/package.json autoctx dependency ^0.11.0 != manifest ^0.9.0" in issues

autocontext/uv.lock

@@ -1,9 +1,9 @@
 {
-  "core_version": "0.13.0",
+  "core_version": "0.14.0",
   "pi_version": "0.9.0",
   "pi_autoctx_dependency": "^0.11.0",
   "whats_new": [
-    "**Python safe active-run stopping** brings `safe_run_stop_v1` to the Python interactive server: stop a running or paused run at a cooperative generation boundary, preserve completed generations and best score, and receive one correlated terminal receipt (first terminal outcome wins).",
-    "**Leaner TypeScript install:** the OpenAI and Anthropic SDKs are now optional peer dependencies, so installing `autoctx` no longer pulls them in transitively. Install the provider SDK you use, for example `npm install autoctx @anthropic-ai/sdk`."
+    "**Live TypeScript task plans:** interactive runs now advertise `agent_task_plan_v1` and emit semantic `task_plan_updated` snapshots for initial planning, meaningful progress, replanning, completion, failure, and safe stop.",
+    "**Durable, privacy-safe replay:** task-plan snapshots use stable identity and monotonic revisions, redact credential-shaped values, reject malformed or oversized frames atomically, and restore exactly across reconnects and server restarts. Python producer parity remains explicitly deferred."
   ]
 }