Native DSH plugin (installed via the official dsh CLI) that runs one-shot or recurring coding tasks in fresh root Agent sessions, manageable from DSH Web or any eligible root Agent, with durable run history, per-automation model targets and explicit `read-only` / `workspace-write` permission boundaries. MIT licensed.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-08-27
License
MIT
01What can it help you accomplish?
Run saved, self-contained coding tasks on a schedule — one-shot, fixed-interval, daily or weekly — inside DeepSeek Harness
Each dispatch starts a fresh root Agent and Session (saved prompt, captured workspace / cwd / Agent preset / permission preset, durable model target) and produces a terminal run record with a result Session ID, bounded summary and structured error
Developers who use DeepSeek Harness (dsh) and need recurring or one-shot coding work to run later without an old chat session
Inspect what each unattended run executed, which revision and model target it used, and how it ended
Durable run history with queued / running / succeeded / failed / skipped / cancelled states, definition revision, prompt and target snapshot, scheduled time, result Session ID, summary and structured error; opt-in archiving of terminal run Sessions
Teams that need unattended coding runs to stay inspectable and auditable
02How to install into DeepSeek Harness
Prerequisites
- A DSH Web profile — the GitHub bundle is installed into the DSH Web profile and `dsh web` is restarted
- The DSH Host must be running for a task to start
- For a local-checkout install: Node.js 22.19 or newer
Installation steps
- 01
Run `dsh plugin --profile web add github:titanwings/dsh-automation#v0.1.7` to install the GitHub bundle into the DSH Web profile (the version tag keeps the install reproducible; a reviewed commit SHA is equally valid)
$ dsh plugin --profile web add github:titanwings/dsh-automation#v0.1.7
- 02
Restart `dsh web`
$ dsh web
- 03
If you run DSH from its source checkout, use `pnpm dsh` in place of `dsh`; for a local checkout run `git clone https://github.com/titanwings/dsh-automation.git`, `cd dsh-automation && pnpm install && pnpm check`, then `pnpm dsh plugin --profile web add /absolute/path/to/dsh-automation`
$ dsh
Verify the integration
- Use **Run now** once before relying on the schedule, and inspect the resulting Session and run record
03DSH integration and capability boundaries
Native DSH/Cordis plugin installed into the DSH Web profile via the official dsh CLI (`dsh plugin --profile web add …`); dispatched runs start in fresh root Agent Sessions inside DSH, with no separate bot, daemon UI, or third-party scheduler.
Schedules people can read (one-shot / interval / daily / weekly)
self-contained task, schedule, IANA time zone, model target and permission boundary→friendly schedule normalized into a validated RFC 5545 RRULE for persistence and inspection
One control plane, two ways in (DSH Web + any eligible root Agent)
DSH Web Automations panel, or natural-language requests to an eligible root Agent→create / pause / resume / run-now / delete rules and inspect recent runs; six scoped Agent tools (automation_create, automation_list, automation_update, automation_run_now, automation_runs, automation_delete)
Per-automation model target
follow the live global model or pin one provider/model pair with optional reasoning effort→each run keeps the same model target in its immutable snapshot
Durable run history with retention
terminal run records (succeeded / failed / skipped / cancelled)→historyLimit (default 200) terminal runs retained per automation; queued and running records never pruned
with archiveRunSessions enabled, terminal run Sessions are archived out of the ordinary DSH conversation list
04Who is it for? When not to use it?
Good for
- Developers who use DeepSeek Harness (dsh) and need recurring or one-shot coding work to run later without an old chat session
- Teams that need unattended coding runs to stay inspectable and auditable
Not for
- Unattended runs receive no inherited authority and only two permission modes are accepted: `read-only` and `workspace-write`; unattended `danger-full-access` is not accepted. Each fresh Session uses approval policy `never` (fail closed), and Agent tools bind to the caller's canonical workspace only.
- Version 0.1 does not provide same-chat heartbeats, raw cron or arbitrary shell actions, unattended full access, automatic retry, Git worktree creation / cleanup, multi-workspace targets / DAGs / hidden cross-run memory, external email / SMS / push delivery, or a guarantee of exactly-once external side effects. Only local execution is implemented.
05Compatibility, maintenance and safety notes
- Unattended runs receive no inherited authority and only two permission modes are accepted: `read-only` and `workspace-write`; unattended `danger-full-access` is not accepted. Each fresh Session uses approval policy `never` (fail closed), and Agent tools bind to the caller's canonical workspace only.
- Minimum interval is five minutes and the first run occurs after one full interval; daily / weekly times are evaluated in an explicit IANA zone with nonexistent DST wall times skipped; a run is cancelled after 60 minutes by default; dispatch is at-most-once, not exactly-once.
- The DSH Host must be running for a task to start; version 0.1 is not an OS daemon and does not coordinate multiple Hosts over one storage directory.
- Version 0.1 does not provide same-chat heartbeats, raw cron or arbitrary shell actions, unattended full access, automatic retry, Git worktree creation / cleanup, multi-workspace targets / DAGs / hidden cross-run memory, external email / SMS / push delivery, or a guarantee of exactly-once external side effects. Only local execution is implemented.
MIT · actively maintained (latest release v0.1.7, 2026-08-23)
06Frequently asked questions
How do I install dsh-automation?
Install the GitHub bundle into the DSH Web profile, then restart `dsh web`. The command is `dsh plugin --profile web add github:titanwings/dsh-automation#v0.1.7`; the version tag keeps the install reproducible, and a reviewed commit SHA is equally valid. If you run DSH from its source checkout, use `pnpm dsh` instead of `dsh`.
How is it different from DSH Core Schedule?
DSH Core Schedule returns to the same live Agent for reminders in the current conversation; dsh-automation starts a fresh root Agent and Session with a saved, self-contained task, a single canonical DSH workspace scope, and durable run records — best for repeated or one-shot standalone coding work.
Can I create automations from an Agent conversation?
Yes. Once installed, eligible root Agents receive six scoped tools — automation_create, automation_list, automation_update, automation_run_now, automation_runs and automation_delete — that manage automations only for the Agent's exact workspace. Plugin-level approval asks for human confirmation when an Agent creates or expands unattended future work.
Is unattended full access supported?
No. Rules may use `read-only` or `workspace-write` only; unattended `danger-full-access` is not accepted. Each fresh Session uses approval policy `never` (fail closed), runs receive no inherited authority, and there are no automatic retries once side effects may have occurred.
What are the current limits of v0.1?
Version 0.1 does not provide same-chat heartbeats, raw cron or arbitrary shell actions, Git worktree creation or cleanup, multi-workspace targets or DAGs, external email / SMS / push delivery, or a guarantee of exactly-once external side effects. Only local execution is implemented, and the DSH Host must be running for a task to start.
07Related DSH workflows
archify
by tt-a1i
Agent skill for beautiful, verifiable architecture, workflow, sequence, data-flow, and lifecycle diagrams—self-contained HTML with motion and crisp export.
openviking
by volcengine
Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
nocobase
by nocobase
NocoBase is an open-source AI + no-code platform for building business systems fast. Instead of generating everything from scratch, AI works on top of production-proven infrastructure and a WYSIWYG no-code interface, so you get both speed and reliability.
learn-harness-engineering
by walkinglabs
Harness engineering beginner tutorial, from 0 to 1
08Data and sources
Install the GitHub bundle into the DSH Web profile, then restart `dsh web`:
The implementation is native to DSH and Cordis; it does not copy Codex internals or patch DSH Core.
There is no separate bot, daemon UI, or third-party scheduler to operate.
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-27. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.