memorix 1.1.13 → 1.2.1

package/CHANGELOG.md

@@ -2,7 +2,39 @@
 
 All notable changes to this project will be documented in this file.
 
-## [Unreleased]
+## [1.2.1] - 2026-07-19
+
+### Added
+- **Review-gated Knowledge claims** -- Explicit agent observations now become source-backed candidates first. They can be inspected and deliberately approved or rejected through `memorix knowledge claims` / `memorix knowledge review` or the advanced `memorix_knowledge` MCP action before they enter knowledge compilation.
+- **Versioned Memorix release workflow** -- Added `docs/knowledge/workflows/memorix-release.md`, a canonical release playbook with verification gates and an explicit maintainer approval boundary.
+
+### Fixed
+- **Claude setup respects `--noHooks`** -- Project setup now keeps generated `CLAUDE.md` guidance without also creating Claude lifecycle-hook configuration when hook capture was explicitly disabled.
+- **Claude Code manual MCP readiness** -- Manual setup examples now include Claude Code's `alwaysLoad: true` entry and point to Doctor/Repair for detecting and restoring the eager-load contract.
+- **Workflow import fidelity** -- Canonical Windsurf workflows preserve their source ID, title, agents, phases, and verification gates instead of being reduced to a generated `workflow:<hash>` entry. Release workflows no longer match a non-release task merely because it says “verify” or “test”.
+- **Graph surface consistency** -- Explicit `relatedEntities` now persist as graph edges, and MCP graph tools, standalone Dashboard, HTTP control plane, and exports share one project-scoping rule.
+- **Intent-aware task and workflow routing** -- A safety constraint such as “do not publish” no longer routes an incident or debugging task into a release lens or release workflow. Explicit release requests still retain release verification while publication is deferred for approval.
+- **Git fact consistency** -- Project Context, Context Pack, and CodeGraph CLI now report `Git: unavailable` for an invalid or unreadable repository instead of presenting it as a clean worktree.
+- **Windows verification-gate timeout** -- A timed-out orchestration gate now resolves promptly while its shell process tree is terminated in the background, instead of waiting indefinitely for a descendant process to close.
+
+## [1.2.0] - 2026-07-18
+
+### Added
+- **Versioned Code State** -- Added local code snapshots, source epochs, completeness metadata, and memory-to-code freshness so project memory can be requalified when the checkout changes.
+- **Claim ledger and reviewable knowledge workspace** -- Added source-backed claims with confidence/conflict lifecycle, plus local or explicitly versioned Markdown workspaces that compile proposals before a reviewed page can change.
+- **Canonical project workflows** -- Added import, selection, preview, safe adapter application, and run receipts for project workflows without treating an agent-specific instruction file as the source of truth.
+- **Bounded task Worksets** -- Memory Autopilot and Context Pack now select task-relevant current facts, evidence, knowledge/workflow starts, cautions, and verification instead of adding a generic historical-memory dump.
+- **Advanced Knowledge Workspace MCP tool** -- Added `memorix_knowledge` as one action-based management surface for workspace, proposal, and workflow operations in the explicit `team` and `full` MCP profiles.
+- **Optional local semantic CodeGraph provider** -- A healthy pre-indexed local CodeGraph can add a validated, bounded semantic outline to a task Workset. The built-in Lite provider remains available with explicit capability limits.
+
+### Changed
+- **Durable knowledge lifecycle** -- Code State refresh, claim derivation/requalification, knowledge compile/lint, and workflow indexing run through resumable maintenance jobs rather than turning interactive requests into a corpus-wide foreground scan.
+- **Provider quality is visible** -- CodeGraph status, Doctor, and Project Context JSON report whether the task used Lite structural evidence or a validated external semantic outline.
+
+### Fixed
+- **Incomplete scan truthfulness** -- Unreadable files and deferred removals now remain visible in snapshot completeness instead of being silently treated as a complete code view.
+- **External CodeGraph safety boundary** -- Memorix only accepts local output for the exact healthy project root, rejects stale/malformed/oversized/path-escaping data, runs without a shell, and never persists raw external source output.
+- **Node 26 SQLite runtime** -- Upgraded the optional `better-sqlite3` path to a release that supports Node 26, and added a Node 26 CI smoke that opens an in-memory database. Fixes #130; reported by @RaviTharuma.
 
 ## [1.1.13] - 2026-07-17
 

package/README.md

@@ -59,11 +59,12 @@ Memorix is more than a memory store. It also installs agent integrations, keeps
 
 | Capability | What it does | Entry point |
 | --- | --- | --- |
-| Memory Autopilot | A task-shaped project brief for a new agent session, with start files, trusted memories, cautions, and verification hints | `memorix context --task "..."`, `memorix_project_context` |
+| Memory Autopilot | A bounded task Workset with start files, current memory, source-backed knowledge, workflow starts, cautions, and verification | `memorix context --task "..."`, `memorix_project_context` |
 | Observation Memory | Searchable facts, fixes, gotchas, session summaries, and implementation notes scoped to the current Git project | `memorix memory`, MCP memory tools |
-| Code Memory | File/symbol links and freshness checks so old memories point to current code or show up as suspect/stale | `memorix codegraph`, automatic context refresh |
+| Code State and Code Memory | Versioned local code snapshots, file/symbol links, and freshness checks. The built-in Lite index is always honest about its limits; an already-indexed local CodeGraph can add a bounded semantic outline | `memorix codegraph`, automatic context refresh |
 | Git Memory | Commit-derived engineering facts that answer what changed, where, and why it matters | `memorix ingest commit`, git hook |
 | Reasoning Memory | Design rationale, alternatives, trade-offs, and risks that should survive beyond one chat | `memorix reasoning`, memory formation |
+| Knowledge Workspace | Review-gated source-backed claims, Markdown pages, and canonical project workflows; proposals never overwrite reviewed pages silently | `memorix knowledge`, `memorix knowledge workflow` |
 | Agent setup | One setup path for MCP, rules, hooks, skills, plugins, bundles, or extensions depending on the agent | `memorix setup --agent <agent>` |
 | Agent doctor | Checks whether agent MCP config and guidance are current, then repairs Memorix-owned entries when needed | `memorix doctor agents`, `memorix repair agents` |
 | Hooks and skills | Optional capture from supported agents, plus reusable project skills promoted from durable knowledge | `memorix hooks`, `memorix skills` |
@@ -257,7 +258,7 @@ What it installs depends on the target agent, but the goal is the same: make Mem
 - Hermes Agent: installs into Hermes home (`%LOCALAPPDATA%\hermes` on native Windows, `~/.hermes` elsewhere, or `HERMES_HOME`), enables the plugin in `config.yaml`, registers plugin hooks, slash/CLI commands, skills, and writes MCP config.
 - Oh-my-Pi: installs an `omp.extensions` package with extension hook events, a `memorix` command, official skills, and writes MCP config.
 
-Need a quieter install? Add `--noHooks` for targets where setup can control hook capture separately from the host's official package entry.
+Need a quieter install? Add `--noHooks` for targets where setup can control hook capture separately from the host's official package entry. It keeps MCP and guidance, but skips Memorix hook capture.
 
 If you intentionally want repo-local guidance or hooks, run the same command inside that repository without `--global`.
 
@@ -274,6 +275,8 @@ If your agent only needs a manual MCP entry, use stdio:
 }
 ```
 
+For a manually managed Claude Code entry, add `"alwaysLoad": true` inside the `memorix` server object. This lets Claude Code expose Memorix tools during print-mode startup; `memorix doctor agents --agent claude` can detect and repair a missing setting.
+
 HTTP is not required for normal setup. Use it only when you intentionally want a shared background service, dashboard, Docker, or multiple clients using the same endpoint:
 
 ```bash

package/README.zh-CN.md

@@ -59,11 +59,12 @@ Memorix 不只是一个记忆库。它还负责安装 Agent 接入、保留有
 
 | 能力 | 作用 | 入口 |
 | --- | --- | --- |
-| Memory Autopilot | 给新 Agent session 一份按当前任务整理的项目 brief，包含起步文件、可信记忆、风险提示和验证建议 | `memorix context --task "..."`、`memorix_project_context` |
+| Memory Autopilot | 给新 Agent session 一份有预算的任务 Workset，包含起步文件、当前记忆、来源知识、工作流首步、风险提示和验证建议 | `memorix context --task "..."`、`memorix_project_context` |
 | Observation Memory | 当前 Git 项目内可检索的事实、修复、坑点、session 摘要和实现记录 | `memorix memory`、MCP memory tools |
-| Code Memory | 文件 / symbol 关联和 freshness 检查；旧记忆会指向当前代码，或标记为 suspect/stale | `memorix codegraph`、自动 context refresh |
+| Code State 和 Code Memory | 可版本化的本地代码快照、文件 / symbol 关联和 freshness 检查。内置 Lite 会如实说明边界；已有本地索引的 CodeGraph 可额外给出有预算的语义关系 | `memorix codegraph`、自动 context refresh |
 | Git Memory | 从 commit 中提取工程事实，回答改了什么、在哪里改、为什么重要 | `memorix ingest commit`、git hook |
 | Reasoning Memory | 保存设计原因、备选方案、trade-off 和风险，不让决策只留在一次聊天里 | `memorix reasoning`、memory formation |
+| Knowledge Workspace | 有审核门槛的来源证据 Claim、Markdown 知识页和规范化项目工作流；提案不会悄悄覆盖已审阅页面 | `memorix knowledge`、`memorix knowledge workflow` |
 | Agent setup | 按目标 Agent 写入 MCP、rules、hooks、skills、plugin、bundle 或 extension | `memorix setup --agent <agent>` |
 | Agent doctor | 检查 Agent 的 MCP 配置和规则是否是当前版本，并修复 Memorix 自己管理的条目 | `memorix doctor agents`、`memorix repair agents` |
 | Hooks 和 skills | 在支持的 Agent 中可选捕获工作事件，并把稳定知识提升成可复用项目技能 | `memorix hooks`、`memorix skills` |
@@ -257,7 +258,7 @@ memorix setup --agent omp --global
 - Hermes Agent：安装到 Hermes home（Windows native 默认为 `%LOCALAPPDATA%\hermes`，其他平台默认为 `~/.hermes`，也支持 `HERMES_HOME`），在 `config.yaml` 启用插件，注册 plugin hooks、slash/CLI commands、skills，并写入 MCP 配置。
 - Oh-my-Pi：安装 `omp.extensions` package，包含 extension hook 事件、`memorix` command、官方 skills，并写入 MCP 配置。
 
-如果你想要更安静一点的安装，可以对那些 setup 能独立控制 hook capture 的 target 加 `--noHooks`。
+如果你想要更安静一点的安装，可以对那些 setup 能独立控制 hook capture 的 target 加 `--noHooks`。它会保留 MCP 和使用规范，只跳过 Memorix 的 hook 自动捕获。
 
 如果你明确想要项目级指导或 hooks，就在那个仓库目录里再跑一次不带 `--global` 的 `memorix setup --agent <agent>`。
 
@@ -274,6 +275,8 @@ memorix setup --agent omp --global
 }
 ```
 
+如果是手动维护 Claude Code 的 MCP 配置，需要在 `memorix` server 对象里加上 `"alwaysLoad": true`。这样 Claude Code 在 print-mode 启动时就会暴露 Memorix tools；`memorix doctor agents --agent claude` 可以检查并修复缺失的设置。
+
 普通安装不需要 HTTP。只有在你明确需要共享后台服务、Dashboard、Docker，或多个客户端共用一个端点时才使用：
 
 ```bash