dsh-missher-memory 0.3.1

dsh-missher-memory 0.3.1

提供预构建的独立 Harness Bundle，适配官方 Harness 0.1.5-rc.2 / Cordis 4.0.2。

安装

dsh plugin --profile web add https://github.com/Missher12/dsh-missher-memory/releases/download/v0.3.1/dsh-missher-memory-0.3.1.tgz

使用目标 Harness 的 CLI；Desktop 用户按仓库安装指南选择其内置运行时。核对附件 SHA-256，重启目标宿主并验证实际插件入口。

GitHub 安装包与市场收录分别维护；目录更新尚待独立审核。现有数据保持原样；本次没有新增三端 Desktop 原生或真实账户验收。已有冻结产物和针对当前变动的检查按其实际范围复用。

SHA-256: 5bde1f688d6791954d890e2b958775abbdaffe06532464c6b23de362cb06ed49

源码提交：a6e462a10a9bd1f3ec2dfe19562d060cd412afbf。

dsh-missher-memory 0.3.0-cordis.0 — Cordis prerelease

Cordis memory prerelease with a portable service and the existing DeepSeek Harness Bundle.

• dsh-missher-memory/core provides project-scoped search, source lookup and review candidates without Harness services. Compatible Agent hosts must map the service to their own tools.
• The Harness adapter retains memory_search, candidate review and settings. Brain is optional for automatic recall; memory content is historical data and grants no authority.
• Includes project isolation, source tracking, reversible duplicate consolidation, schema 2 migration, corruption guards, and data retention across uninstall/reinstall.
• Fixes derived-memory cleanup when forgetting a source, capsule search and budget handling, duplicate matching, and project deletion cleanup.

This is a prerelease, not a claim that arbitrary Agents work without integration. Requires Node ^22.19.0 || >=24.0.0. Real model tool invocation and Desktop Brain/settings UI are not accepted by the synthetic checks. Forgotten candidate text remains retained; project JSON export is not a full backup.

Install the prebuilt Bundle:

dsh plugin --profile web add https://github.com/Missher12/dsh-missher-memory/releases/download/v0.3.0-cordis.0/dsh-missher-memory-0.3.0-cordis.0.tgz
dsh --profile web --dump-config

Restart the compatible host, explicitly bind the project, and confirm that the current Agent can call memory_search. Configuration parsing alone does not prove activation. For other Cordis hosts, follow CORDIS.md in the package/repository.

Source and verification details are recorded in the attached release-verification.json. The .tgz.sha256 file covers the exact prebuilt package. All acceptance data is synthetic and uses isolated homes.

商店收录需要上游维护者审核；GitHub 发布成功与商店可搜索是两个独立状态。

Verification: 121 tests / 29 files; typecheck, package checks, independent Cordis lifecycle and native CLI install/uninstall/reinstall passed on macOS Intel, macOS Apple Silicon, Windows x64 and Linux x64. All four runners tested the same archive.

Source SHA: 8a8c796b75f08995b7af7a49be9e0c5a5bc08fd4

Package SHA-256: 6b1a8a4fe4b5e90e94e9c6ef3406fa1347551efee45a8c352d5e0b57d810d886

Cross-platform CI

dsh-missher-memory v0.2.0

Project-isolated reviewed memory for DeepSeek Harness.

• Schema v2 with transactional migration and reversible consolidation capsules
• Local FTS5 search with CJK bigrams and fixed literal queries
• Bundled read-only TencentDB compatibility reader (reader code only; no user database is packaged)
• One provider path through dsh-missher-brain; no duplicate prompt injection
• Cross-platform package and native lifecycle verification on macOS Intel/Apple Silicon, Windows, and Linux

SHA-256: 4790bf908c5c671b8a77029e62cea545bba80212d24acc02b9ddea90f527bc03

dsh-missher-memory 0.1.3

Polishes the Project Memory settings copy: removes platform-specific wording and internal database filenames from customer-facing text. Adds a regression test that prevents those implementation details from returning. The memory model, project isolation, local capture/recall behavior, and existing data remain unchanged.

dsh-missher-memory v0.1.2

Fresh Windows and macOS installs now show built-in project memory as ready without requiring a legacy vectors.db.

The optional legacy vector index remains read-only. Project memory still requires explicit project binding and preserves existing privacy and approval boundaries.

Validated on macOS Intel, macOS Apple Silicon, Linux x64, and Windows x64 in CI run 32680947534.

dsh-missher-memory v0.1.1

Reviewed local memory for DeepSeek Harness long-running projects.

• default-on candidate capture and bounded recall for newly bound projects
• project isolation, source visibility, and review-before-searchable storage
• read-only TencentDB discovery with path and sensitive-data guards
• native acceptance on macOS Intel, macOS Apple Silicon, Windows x64, and Linux x64

Canonical package SHA-256: 4ad84c3f4ea73dcab06076185bd9e55c0677b5a246033b76df7bebb1203121e1