Workflow & Automation

Read-only orchd panel for the DSH right sidebar: task board, ideas board, dependency graph, review timeline and an install entry for the project of the current session.

Arms a watch on a directory and delivers a batched, deduplicated diff when the tree goes quiet, filtering out the writes the session made itself; delivery is per watch, either waking a new turn or queueing for the next step.

Zagent — unified three-role self-improvement loop with three operators. Data-RSI (daily cross-sectional cleaning: winsorize → cross-sectional standardization → median fill, plus experience extraction), Harness-RSI (evolves system prompt / memory / skills), and Model-RSI (evolves alpha training recipes via framework-native walk-forward with 2-day gap and multi-seed). Offline one-command run, GPU auto-detect. Ships one skill.

Host-side long-task detection for small-context local models: long outputs are scaffolded into a parts skeleton (one part per reply, then assemble), and multi-step work into a PLAN.md ledger with notes/ (plan first, one step per reply). Detection is heuristic and host-side, so it does not depend on the model judging its own limits.

Verified-progress ledger: three tools record a claimed step as progress only after an independent subagent with a fresh context confirms it against the workspace. The declared scope is fingerprinted before and after each verification, so a verifier that changed anything has its verdict voided and the round cannot become progress.

Persistent Bots and group rooms inside DSH. Each Bot keeps its own model, permission preset, workspace and session across restarts; a Bot can also be backed by an installed local agent CLI (Codex CLI, Claude Code, ACP agents) using that CLI's own login and models. Unaddressed messages can be triaged by the room leader, the room accumulates a shared experience card, and every answer is a run in a durable ledger the sidebar panel replays. New Bots default to the full-access permission (any command, any file; the form asks you to confirm the risk), so choose read-only or workspace-write to confine one; Codex CLI and Claude Code Bots at full access run without their own sandbox or permission prompts.

A resident assistant for DeepSeek Harness that is deliberately not another chat box: one floating window plus one standing conversation, both independent of every other session. The heartbeat is held back by four gates — an authority grant, a daily turn budget, quiet hours and a pacing cooldown — and it keeps working across restarts through a lease it has to hold. Each wake is decided by a model that must answer SKIP or a short message, so silence is a decision rather than a failure. The window shows the standing conversation's real message flow, read from the host session store, or from the on-disk session log when the host does not publish that session, and it can send into that conversation directly. den carries the same reach as the person it stands in for, with one limit: read-only. It may read every session, this machine and work channels such as GitHub, but it does not write, send or publish, it never asks anyone for permission, and it never gates another session's tool calls. `npm run digest:github` turns GitHub notifications and your own open pull requests into three Chinese sections — waiting on you, waiting on your review, and yours — with reply drafts that are explicitly marked as unsent.

Daily brief for DeepSeek Harness. Every day at a fixed hour it ranks the plugin market by real install counts and stars, cross-checks it against your own agent memory and working conventions, and proposes two kinds of item: concrete deltas on the plugins people actually use (original does A, the proposal does B) and directions nobody has built yet. The popularity numbers come from the catalog the market itself publishes and are quoted into the prompt rather than guessed; an entry the market has not measured is shown as unmeasured, never as zero. Approving an item opens a new session named after it and hands that session a fixed delivery contract — implement independently, never fork, test green before publishing. Rejected sources are remembered so they stop coming back.

Daily and weekly task manager: startup popup reminder, a marquee strip above the composer, one-click circle completion, editable task cards, and a Sunday AI weekly summary, plus a sidebar DeepSeek API usage widget with balance card, today/cumulative tokens and cost, per-model breakdown and a 7-day chart. Persists as JSON under $DSH_HOME.

Multi-agent delivery workflow for dsh: a commander dispatches to fixed and temporary agents through a persisted task state machine, with a canvas and roster in the right sidebar.

Task workspace convention: one folder per session, one PROGRESS.md per task overwritten on each update, and per-task git once the task runs long.

Gated four-stage video pipeline: brief, script, assets and compose, turning one sentence into a narrated explainer with ComfyUI workflows or image/video APIs and FFmpeg assembly.

Session-scoped research evidence ledger with sources, linked claims, open questions, and explicit user review; provides a tool, slash command, and conversation panel, with selected-claim Markdown export that omits excerpts, review notes, and local paths by default; does not fetch sources or verify their truth.

Session-scoped delivery receipts for local artifacts: compare current file hashes with a recorded baseline, keep agent claims separate from explicit user acceptance, flag changed or unavailable files, and export a Markdown receipt through a tool, slash command, and conversation panel.

Ports jcode's auto-poke so an idle agent keeps working while its session still has incomplete todos.

Recurring Excel template fill and reconciliation review: declare the file, sheet, header row, match key and amount columns once, preview every cell write before it lands, compare amounts as exact two-decimal values, list the unmatched rows as exceptions, and export a filled template copy with an HTML report and a JSON trace back to the source rows.

Declares a spec-driven development agent preset with a Chinese persona and four static Claude Code/Codex delegation tools for planning, implementation, review and knowledge distillation; roles are prompt-guided, not an automatic state machine.

Thesis and empirical-analysis workbench that drives IBM SPSS from the conversation: spss_check (environment self-check), spss_variables (list .sav variables with labels, types, measurement levels, case counts), spss_data (sample case rows to inspect missing and outlying values), spss_run / spss_run_file (execute SPSS syntax or an .sps file, relative CD and GET FILE paths auto-rebased), and spss_analyze (one-shot frequencies, descriptives, crosstabs with chi-square, t-tests, one-way ANOVA, correlation, linear regression). Results come back as structured tables extracted through OMS. SPSS 25-31 is auto-detected; the bundled MCP server is zero-dependency stdio.

Lets Codex delegate tasks to Desktop sessions through a local authenticated bridge and companion MCP server, with Workspace grouping, model and reasoning effort selection, event tracking, waiting, and cancellation.

Persistent task supervision with plan coverage review, DAG progress, sidebar consultation, and stage/completion checks; npm development preview.

面向 DSH（DeepSeek Harness）的专家与专家团插件：给 AI 加"角色与团队"两个协作维度——用一个领域专家的身份干活，或拉起一支多 Agent 团队按流程交付。

A Codex delegation tool for DeepSeek Harness that can carry a model and a reasoning effort per call.

dsh plugin: Jev-guided dispatch advice — before a turn starts, a TypeSafe Jev call decides whether the task belongs to a registered agent and which one, then injects a four-line, ignore-if-unsuited recommendation. Jev-optional: no key means silence.