chore: prepare v0.19.0 release

Author: codejunkie99

CHANGELOG.md

@@ -5,6 +5,34 @@ All notable changes to this project.
 The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and the project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.19.0] — 2026-07-18
+
+Adds portable, bounded agentic loops and a local meta-harness for resumable
+maker/verifier/checker work.
+
+### Added
+- Strict JSON loop contracts, executor profiles, constraints, budgets, and
+  portable seed skills.
+- Atomic checkpoints, privacy-safe events, shell-free bounded subprocesses,
+  owned Git worktrees, policy gates, stagnation breakers, and CLI lifecycle
+  verbs (`init`, `validate`, `run`, `resume`, `status`, `stop`, `cleanup`,
+  `audit`).
+- Read-only loop health in doctor/status/dashboard/Mission Control and hashed
+  loop-event ingestion in the local data layer.
+
+### Safety
+- L2/L3 action loops require worktree isolation, deterministic verification,
+  independent checker approval, finite budgets, and explicit approval before
+  mutating or external-write profiles.
+- Runtime events omit task/prompt/command/output content by construction. The
+  supervisor is not an operating-system sandbox; use harness-native isolation
+  for hostile child processes.
+
+### Migration
+Run `agentic-stack upgrade --dry-run`, then `agentic-stack upgrade --yes` in
+existing projects. Upgrade adds missing loop starters and skills but preserves
+authored contracts, runtime state, and existing loop skill directories.
+
 ## [0.18.0] — 2026-05-10
 
 Minor release. Adds first-class integration with the external Brain CLI/MCP

README.md

@@ -43,6 +43,24 @@ workspace into agentic-stack.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full list.
 
+### New in v0.19.0 — bounded agentic loops
+
+Portable loop contracts live under `.agent/loops` and use a maker →
+deterministic verifier → independent checker lifecycle. Start with:
+
+```bash
+agentic-stack loop init /path/to/your-project
+agentic-stack loop validate /path/to/your-project
+agentic-stack loop run ci-sweeper "make the failing test green" /path/to/your-project --yes
+agentic-stack loop status /path/to/your-project
+```
+
+L2/L3 action loops use owned Git worktrees, finite attempts/runtime/output/token
+budgets, deny-path gates, resumable checkpoints, and privacy-safe local events.
+The supervisor bounds and audits child processes; it is not an operating-system sandbox. Use harness-native sandboxes and approvals for stronger isolation.
+Schedulers should invoke one bounded `loop run` command at a time and inspect
+its exit status before starting another run.
+
 ### v0.17.0 — adapters, Mission Control, and lesson retraction
 
 Minor release. Clears the open PR queue and ships the combined production

docs/getting-started.md

@@ -89,6 +89,26 @@ agentic-stack upgrade --yes       # apply latest harness/memory/tools + new skil
 agentic-stack sync-manifest       # rebuild .agent/skills/_manifest.jsonl from SKILL.md
 ```
 
+### Bounded agentic loops
+
+Initialize and inspect the portable loop contracts before running an action loop:
+
+```bash
+agentic-stack loop init /path/to/your-project
+agentic-stack loop validate /path/to/your-project
+agentic-stack loop run ci-sweeper "make the failing test green" /path/to/your-project --yes
+agentic-stack loop status /path/to/your-project
+```
+
+The lifecycle is maker → deterministic verifier → independent checker. L1
+loops report in the current workspace; bundled L2/L3 action loops use owned
+worktrees, approvals, finite budgets, deny-path gates, and resumable local
+checkpoints. Events contain only allowlisted metadata and hashed identifiers.
+The loop supervisor is **not an operating-system sandbox**; pair it with a
+harness-native sandbox when hostile or high-impact child code is in scope.
+Schedulers should run one bounded command, inspect its exit code, and only then
+schedule the next run.
+
 Source checkout users can run the same verbs through the clone:
 
 ```bash

docs/releases/v0.19.0.md

@@ -0,0 +1,45 @@
+# agentic-stack v0.19.0
+
+v0.19.0 adds portable agentic loops and a bounded local meta-harness.
+
+## Highlights
+
+- JSON contracts for L1 report loops and L2/L3 action loops.
+- Maker → deterministic verifier → independent checker feedback lifecycle.
+- Atomic resumable checkpoints, privacy-safe local events, policy gates,
+  stagnation detection, finite budgets, and owned Git worktrees.
+- `loop init`, `validate`, `run`, `resume`, `status`, `stop`, `cleanup`, and
+  `audit` commands through both `agentic-stack` and `./install.sh`.
+- Read-only loop summaries in doctor, status, dashboard, Mission Control, and
+  the local data-layer export.
+
+## Usage
+
+```bash
+agentic-stack loop init /path/to/project
+agentic-stack loop validate /path/to/project
+agentic-stack loop run ci-sweeper "make the failing test green" /path/to/project --yes
+agentic-stack loop status /path/to/project
+```
+
+Use `agentic-stack upgrade --dry-run` followed by `agentic-stack upgrade --yes`
+to add loop starters to an existing installed project without overwriting
+authored contracts or runtime state.
+
+## Safety boundaries
+
+L2/L3 loops run mutating makers in owned worktrees, enforce deny-path and file
+count gates, require deterministic verification and independent checker
+approval, and stop on pause, budget, timeout, or repeated-failure decisions.
+Events are local-only and whitelist metadata; prompts, tasks, command arguments,
+stdout, stderr, and raw run IDs are not exported. The supervisor is not an operating-system sandbox; use harness-native sandboxes for stronger isolation.
+Schedulers should execute one bounded command and inspect its exit status before
+starting another.
+
+## Verification and attribution
+
+The release is covered by the focused loop, integration, and legacy-hook suites;
+the final release commit records the fresh full-suite command and count. The
+loop design was informed by public loop-engineering discussions about bounded
+feedback, verifier gates, resumability, and stagnation breakers. This project
+implements its own Python standard-library contracts and does not copy code.

harness_manager/__init__.py

@@ -6,4 +6,4 @@
 The "harness_manager" name is internal only and never appears in CLI help, docs,
 or error messages users see.
 """
-__version__ = "0.18.0"
+__version__ = "0.19.0"

tests/test_onboard_ui.py

@@ -1,7 +1,22 @@
 from harness_manager import __version__
 import onboard_ui
+from pathlib import Path
 
 
 def test_onboarding_banner_uses_package_version():
     assert f"v{__version__}" in onboard_ui._T
     assert "v0.8.0" not in onboard_ui._T
+
+
+def test_v019_release_docs_describe_loops_and_safety():
+    root = Path(__file__).resolve().parents[1]
+    assert __version__ == "0.19.0"
+    readme = (root / "README.md").read_text(encoding="utf-8")
+    getting_started = (root / "docs" / "getting-started.md").read_text(encoding="utf-8")
+    changelog = (root / "CHANGELOG.md").read_text(encoding="utf-8")
+    release = (root / "docs" / "releases" / "v0.19.0.md").read_text(encoding="utf-8")
+    for text in (readme, getting_started, release):
+        assert "loop init" in text and "loop run" in text and "loop status" in text
+        assert "not an operating-system sandbox" in text
+    assert "## [0.19.0]" in changelog
+    assert "### Added" in changelog and "### Safety" in changelog and "### Migration" in changelog