Release Helm 0.9.5

Author: kevin

CHANGELOG.md

@@ -11,6 +11,23 @@
 - **PyPI publish hardening**: the publish workflow now runs `twine check` before publishing release artifacts.
 - **release smoke isolation**: package install verification now uses a smoke-run virtual environment instead of writing to the active user Python environment.
 
+## [0.9.5] — 2026-05-15
+
+### Added
+
+- **OpenHuman-inspired operations digest boundary**: documented the reusable pattern from OpenClaw dogfooding: artifact fingerprints, connector freshness, daily digest review queues, and recoverable JSONL retention.
+- **OpenClaw integration guidance**: expanded the integration doc with a public-safe boundary for promoting operations metadata without private memory, schedules, credentials, or raw task content.
+
+### Changed
+
+- **README positioning**: added operations digest and connector freshness to the core Helm value proposition in English and Korean.
+- **release metadata**: bumped package, citation, README, changelog, and release-note metadata to `0.9.5`.
+
+### Verification
+
+- `python3 scripts/release_version_check.py --version 0.9.5`
+- `python3 -m pytest -q`
+
 ## [0.9.4] — 2026-05-13
 
 ### Added

CITATION.cff

@@ -8,7 +8,7 @@ authors:
 repository-code: "https://github.com/JDeun/Helm"
 url: "https://v0-helm-agent-ops.vercel.app/"
 license: "MIT"
-version: "0.9.4"
+version: "0.9.5"
 abstract: "Helm is a local operations layer for long-lived AI agent workspaces, adding execution profiles, guard decisions, checkpoints, task ledgers, context hydration, and recovery discipline around existing agent runtimes."
 keywords:
   - ai-agents

README.ko.md

@@ -8,7 +8,7 @@
 
 <p align="center">Helm은 AI agent workspace를 위한 로컬 운영 레이어입니다. 명령 전 profile, 위험 작업 전 checkpoint, chat이 사라진 뒤에도 남는 durable task history를 제공합니다.</p>
 
-<p align="center"><strong>현재 릴리즈: v0.9.4</strong></p>
+<p align="center"><strong>현재 릴리즈: v0.9.5</strong></p>
 
 <p align="center">
   <a href="https://v0-helm-agent-ops.vercel.app/">Landing page</a> ·
@@ -131,6 +131,7 @@ helm dashboard --path ~/.helm/workspace
 - **File-backed memory**: 다음 실행이 chat history가 아니라 파일에 남은 durable state에서 이어지게 합니다.
 - **Context retrieval**: notes, memory, ontology, tasks, commands, checkpoints를 하나의 inspectable query surface로 정렬합니다.
 - **Privacy boundary**: private text가 tool, API, report, remote handoff 경계를 넘기 전에 scan/tokenize합니다.
+- **Operations digest**: capture 상태, artifact fingerprint, connector freshness, review pressure를 private workspace 내용 없이 요약합니다.
 
 | 반복 에이전트 운영 문제 | Helm이 더하는 것 |
 | --- | --- |
@@ -142,6 +143,7 @@ helm dashboard --path ~/.helm/workspace
 | skill 규칙이 프롬프트에만 남음 | `SKILL.md` guidance와 `contract.json` 실행 정책 |
 | model fallback이 즉흥적으로 결정됨 | file-backed health check와 fallback selection |
 | 운영 상태가 흩어짐 | workspace layout, adopted sources, SQLite query index |
+| 장기 연동이 조용히 stale 상태가 됨 | connector freshness probe와 daily digest review queue |
 
 Helm은 원칙적으로 runtime-agnostic이지만, state, memory, profiles, checkpoints, task history가 있는 persistent workspace를 1차 대상으로 설계되었습니다.
 
@@ -250,7 +252,11 @@ Helm은 전용 workspace에 두고, 기존 시스템은 먼저 read-only context
 - [`docs/ops-memory-query.md`](docs/ops-memory-query.md)
 - [`docs/privacy-boundary.md`](docs/privacy-boundary.md)
 - [`docs/task-finalization.md`](docs/task-finalization.md)
+- [`docs/task-state.md`](docs/task-state.md)
+- [`docs/integrations/openclaw.md`](docs/integrations/openclaw.md)
 - [`docs/adaptive-harness.md`](docs/adaptive-harness.md)
+- [`docs/evidence-label-convention.md`](docs/evidence-label-convention.md)
+- [`docs/hitl-decision-patterns.md`](docs/hitl-decision-patterns.md)
 - [`docs/skill-quality-and-policy.md`](docs/skill-quality-and-policy.md)
 
 포지셔닝:
@@ -269,14 +275,14 @@ Helm은 전용 workspace에 두고, 기존 시스템은 먼저 read-only context
 - [`CHANGELOG.md`](CHANGELOG.md)
 - [`CONTRIBUTING.md`](CONTRIBUTING.md)
 - [`SECURITY.md`](SECURITY.md)
-- 최신: [`docs/releases/0.9.4.md`](docs/releases/0.9.4.md)
-- 최근: [`0.9.3`](docs/releases/0.9.3.md), [`0.9.2`](docs/releases/0.9.2.md)
+- 최신: [`docs/releases/0.9.5.md`](docs/releases/0.9.5.md)
+- 최근: [`0.9.4`](docs/releases/0.9.4.md), [`0.9.3`](docs/releases/0.9.3.md)
 
 이전 릴리즈 노트는 [`docs/releases/`](docs/releases/)에 있습니다.
 
 ## 현재 상태
 
-Helm v0.9.4는 release smoke 패키징 검증과 단일 패키지 버전 source를 추가합니다. 자세한 내용은 [`docs/releases/0.9.4.md`](docs/releases/0.9.4.md) 참고.
+Helm v0.9.5는 OpenHuman에서 차용한 operations digest 경계를 문서화하고 release metadata를 동기화합니다. 자세한 내용은 [`docs/releases/0.9.5.md`](docs/releases/0.9.5.md) 참고.
 
 Helm에는 private memory, personal agent overlay, credential, private task history가 포함되지 않습니다.
 

README.md

@@ -8,7 +8,7 @@
 
 <p align="center">Helm is a local operations layer for AI agent workspaces: profiles before commands, checkpoints before risky work, durable task history after the chat is gone.</p>
 
-<p align="center"><strong>Current release: v0.9.4</strong></p>
+<p align="center"><strong>Current release: v0.9.5</strong></p>
 
 <p align="center">
   <a href="https://v0-helm-agent-ops.vercel.app/">Landing page</a> ·
@@ -131,6 +131,7 @@ Core ideas:
 - **File-backed memory**: keeps reusable context in files so later runs resume from durable state instead of chat history.
 - **Context retrieval**: ranks notes, memory, ontology, tasks, commands, and checkpoints through one inspectable query surface.
 - **Privacy boundary**: scans and tokenizes private text before it crosses tool, API, report, or remote handoff boundaries.
+- **Operations digest**: summarizes capture status, artifact fingerprints, connector freshness, and review pressure without exposing private workspace contents.
 
 | Repeated-agent problem | Helm adds |
 | --- | --- |
@@ -142,6 +143,7 @@ Core ideas:
 | Skill rules live in prompts | `SKILL.md` guidance plus `contract.json` execution policy |
 | Model fallback is ad hoc | File-backed health checks and fallback selection |
 | Operational state is scattered | Workspace layout, adopted sources, and SQLite query index |
+| Long-lived integrations silently go stale | Connector freshness probes and daily digest review queues |
 
 Helm is runtime-agnostic, but it is built first for persistent workspaces with state, memory, profiles, checkpoints, and task history.
 
@@ -253,6 +255,7 @@ Core concepts:
 - [`docs/privacy-boundary.md`](docs/privacy-boundary.md)
 - [`docs/task-finalization.md`](docs/task-finalization.md)
 - [`docs/task-state.md`](docs/task-state.md)
+- [`docs/integrations/openclaw.md`](docs/integrations/openclaw.md)
 - [`docs/adaptive-harness.md`](docs/adaptive-harness.md)
 - [`docs/evidence-label-convention.md`](docs/evidence-label-convention.md)
 - [`docs/hitl-decision-patterns.md`](docs/hitl-decision-patterns.md)
@@ -274,14 +277,14 @@ Release details:
 - [`CHANGELOG.md`](CHANGELOG.md)
 - [`CONTRIBUTING.md`](CONTRIBUTING.md)
 - [`SECURITY.md`](SECURITY.md)
-- Latest: [`docs/releases/0.9.4.md`](docs/releases/0.9.4.md)
-- Recent: [`0.9.3`](docs/releases/0.9.3.md), [`0.9.2`](docs/releases/0.9.2.md)
+- Latest: [`docs/releases/0.9.5.md`](docs/releases/0.9.5.md)
+- Recent: [`0.9.4`](docs/releases/0.9.4.md), [`0.9.3`](docs/releases/0.9.3.md)
 
 Older release notes live in [`docs/releases/`](docs/releases/).
 
 ## Status
 
-Helm v0.9.4 adds release-smoke packaging validation and a single package version source. See [`docs/releases/0.9.4.md`](docs/releases/0.9.4.md).
+Helm v0.9.5 documents the OpenHuman-inspired operations digest boundary and keeps release metadata in sync. See [`docs/releases/0.9.5.md`](docs/releases/0.9.5.md).
 
 Helm does not include private memory, personal agent overlays, credentials, or private task history.
 

docs/integrations/openclaw.md

@@ -28,6 +28,25 @@ helm status --path ~/.openclaw/workspace --brief
 helm report --path ~/.openclaw/workspace --format markdown
 ```
 
+## Operations Digest Boundary
+
+OpenHuman-inspired dogfooding in OpenClaw produced a reusable operations pattern that belongs in Helm as a public-safe primitive:
+
+- artifact registry rows should store path identity, content hash, byte size, role, source task metadata, and duplicate links
+- connector freshness rows should store last attempt, last success, stale threshold, last error, token scope label, and artifact count
+- daily digest output should summarize task status, capture status, artifact roles, connector freshness, and review queue pressure
+- retention policy should compact high-volume JSONL logs into recoverable archives instead of silently deleting history
+- status surfaces should expose summaries and pointers, not raw private memory, personal schedules, credentials, or task content
+
+The OpenClaw-specific part stays private:
+
+- actual connector names tied to personal accounts
+- raw memory captures, notes, schedules, and command outputs
+- local vault paths and private workspace state
+- user-specific automation cadence and notification routing
+
+The portable Helm rule is: preserve evidence and freshness metadata, but keep private content behind the workspace boundary.
+
 ## Promotion rule
 
 Promote to Helm only when a pattern is:

docs/releases/0.9.5.md

@@ -0,0 +1,25 @@
+# Helm v0.9.5
+
+Helm v0.9.5 documents the OpenHuman-inspired operations digest boundary extracted from OpenClaw dogfooding and keeps release metadata synchronized.
+
+## Added
+
+- OpenClaw integration docs now describe the portable operations digest pattern:
+  - artifact registry metadata with content hashes and source task context
+  - connector freshness records with stale thresholds and last-attempt state
+  - daily digest summaries for capture status, artifact roles, connector freshness, and review queue pressure
+  - recoverable JSONL retention through archives instead of silent deletion
+- README and Korean README now include operations digest and connector freshness as first-class Helm concepts.
+
+## Changed
+
+- Bumped package metadata, citation metadata, README release banners, changelog, and release-note links to `0.9.5`.
+
+## Boundary
+
+This release is documentation and version-management only. It does not ship private OpenClaw memory, personal connector state, local paths, credentials, schedules, or raw task history.
+
+## Verification
+
+- `python3 scripts/release_version_check.py --version 0.9.5`
+- `python3 -m pytest -q`

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "helm-agent-ops"
-version = "0.9.4"
+version = "0.9.5"
 description = "Stability-first operations CLI for long-lived agent workspaces."
 readme = "README.md"
 requires-python = ">=3.10"