DSH Skill for organizing DeepSeek Harness workspace sessions as 类别｜主题, with archive/rename/judgment suggestions. DSH & ChatGPT supported; Claude not supported.
Pending human review — the facts below come from the source repository.
DSH Skill for organizing DeepSeek Harness workspace sessions as 类别｜主题, with archive/rename/judgment suggestions. DSH & ChatGPT supported; Claude not supported.
Users who want this capability in DSH; check the source and docs before installing.
dsh plugin --profile web add github:caoqinnan-web/organize-workspace-sessions No supported platforms were declared.
类别｜主题 (Category｜Topic) titles, and report safe-to-archive / rename / needs-judgment suggestions. This Skill organizes the sessions in the current DSH workspace through the host's local RPC interface (workspace.list / session.list / session.history / session.rename). It renames every visible session to 类别｜主题. Because DSH currently has no way to view archived sessions, it does not archive anything — it only lists safe-to-archive candidates in the report. 中文说明 ## Usage 1. Open the DeepSeek Harness workspace you want to organize. 2. Start a new session in that workspace. 3. Say “整理对话” (or “整理会话” / “整理工作区会话” / “清理会话”, etc.). The Skill then inventories the workspace's sessions, renames them as 类别｜主题, and reports archive/rename/judgment suggestions. ## Compatibility | Environment | Status | Notes | |---|---|---| | DeepSeek Harness / DSH | ✅ Supported | Renames workspace sessions via the host's local RPC interface. | | ChatGPT | ✅ Supported | Verified by the author (the same Category｜Topic renaming workflow). | | Claude | ❌ Not supported | Tested by the author; Claude does not expose the required capabilities. | | Other Agent hosts | Capability-dependent | The host must expose session listing, full-content reading, title updates, and result rereading. | ## What the Skill does text 锁定工作区 → 建立清单 → 阅读正文 → 分级 → 改名 → 回读验证 - Locks to the current workspace and reconciles visible sessions against archived ones. - Reads each session's real content (only source.kind === "user" messages count as the user's question). - Classifies each session as A (safe-to-archive), B (needs judgment), or C (keep). - Renames every visible session as 类别｜主题 (exactly one full-width separator). - Because DSH has no archive-view entry, it does not archive; A-class sessions are reported as archive suggestions. - Rereads the session list to verify every rename. It changes session titles only. It does not archive, delete, move, merge, or edit session content. ## How it works The Skill drives the DSH host through its local RPC interface (POST /api/<method> on $DSH_WEB_URL, default http://127.0.0.1:3080): - workspace.list / session.list — inventory - session.history — read real content - session.rename — pin a 类别｜主题 title - workspace.archiveSession — documented but intentionally not called for now It ships two helper scripts: - skills/organize-workspace-sessions/scripts/dsh_rpc.sh — the RPC envelope - skills/organize-workspace-sessions/scripts/session_digest.py — one-pass content digest for fast grading ## Installation As a DSH plugin: bash dsh plugin --profile web add "github:caoqinnan-web/organize-workspace-sessions#main" Or use the skill folder directly: drop skills/organize-workspace-sessions into your skills directory. ## Repository structure - skills/organize-workspace-sessions/SKILL.md — the skill instructions. - skills/organize-workspace-sessions/scripts/ — helper scripts (dsh_rpc.sh, session_digest.py). - skills/organize-workspace-sessions/agents/openai.yaml — interface metadata. - src/, cordis.patch.yml — the DSH plugin wrapper that registers the skill. ## Development and validation bash npm install npm run check ## License MITNot by stars — picked by hand, one at a time.
A visual settings dashboard for DeepSeek Harness: every registered settings namespace — including third-party plugin ones the official UI never covers — rendered as editable, schema-driven forms. Early, but the right direction: stop hand-editing YAML.
A DIY trajectory visualizer for DeepSeek Harness: watch the agent’s plan and execution unfold as a graph instead of scrolling raw logs. Early, but it fills a real gap — understanding at a glance what the agent is doing.
Dispatch DSH work from Claude Code / Codex: spin up in-host DSH agent sessions with per-tier presets, watch native subagent progress, and borrow a multimodal bridge that lends the text-only harness vision and image generation. An orchestration plugin that turns DSH into a backend for other coding agents.