Plugin directory / Developer / dsh-subagent-profile
dsh-subagent-profile
Verified · install-tested on dsh muzyLink
What it does
DeepSeek Harness 子 Agent 派发方案化插件:按任务指定预设/模型/推理强度/工具范围,常用组合存成命名方案一键派发,内置成本护栏。
Works — verified, early-stage project
DeepSeek Harness 子 Agent 派发方案化插件:按任务指定预设/模型/推理强度/工具范围,常用组合存成命名方案一键派发,内置成本护栏。 It installs cleanly and boots without issues in our testing. It's early-stage but functional.
“Verified” means our automated CI actually ran dsh plugin add in a clean profile and it booted — nothing more. Feature descriptions and version compatibility are the author’s claims. This is not a security audit and not an endorsement of third-party code.
README
dsh-subagent-profile
For DeepSeek Harness (DSH).
Thinking, Fast and Slow: System 1 is fast and cheap, System 2 is slow and careful. The built-in
subagentgives every subtask the same brain as its parent — no way to tell them apart.dsh-subagent-profilelets you pick per subtask: research with a fast brain, deep work with a careful one, saved as named profiles.
Why this plugin
Built-in subagent |
dsh-subagent-profile |
|
|---|---|---|
| Per-subtask model / preset | ❌ same brain for every subtask | ✅ pick per subtask |
| Reusable named setups | ❌ | ✅ profiles |
| Tool-scope narrowing | ❌ | ✅ whitelist ∩ parent, run_code always removed |
| Cost guardrails | ❌ | ✅ model / effort / tokens / depth capped |
| GUI management | ❌ | ✅ settings page |
- Per-subtask control over the child's brain.
dispatchsets, per subtask: which preset (composition), which model, which reasoning effort, which tools, and the token cap. A research subtask and a coding subtask can run with completely different setups — something the plainsubagenttool can't do (it only inherits the parent). - Named, reusable profiles. A profile is one bundle of preset + model + reasoning effort + tool scope + persona. Save "research" as
researcher(reasoning off, search-only tools) and dispatch withdispatch(profile="researcher"); two built-ins ship (swap-standard= full standard coding toolkit,researcher), and you can add/edit/remove your own in the settings page. - Fully observable. Every result reports the effective profile / preset / model / reasoning effort; logs are tagged
[dsh-subagent-profile].
Installation
dsh plugin --profile web add dsh-subagent-profile # published package
dsh plugin --profile web add ./dsh-subagent-profile # from a local checkout
Restart dsh web. This is a standard bundle plugin: it provides the dispatch tool, the profile provider, the subagent-profiles service, the /subagent-profiles/* loopback management routes, the settings page (「子 Agent 方案」), and the dispatch tool-call card in the web GUI. On startup it also self-installs an agent preset — orchestrator (「编排者模式」) — pick it in the new-session preset picker. The sync is idempotent and re-runs on every startup, so upgrading the plugin updates the preset.
Usage
1. Configure sub-agent profiles
Profiles are managed in the settings page — each one bundles preset + model + reasoning effort + tool scope (and optionally a persona), and can be enabled, disabled, edited, or reset individually. Two built-ins ship:
| Profile | Purpose |
|---|---|
swap-standard |
switch the child to the full standard coding toolkit |
researcher |
deep reasoning off, search-only tools |
Profiles live in ~/.dsh/subagent-profiles.json and take effect immediately (edits are made from the settings page).


2. Dispatch per subtask — the dispatch tool
dispatch(
profile: "researcher", // preset + model + reasoning effort + tool scope
prompt: "Survey the DSH plugin ecosystem and compare direct competitors",
run_in_background: true
)

Safety model
Delegation never lets a subagent gain more power than you already have — this is the default, with no configuration:
- Tools only shrink. A child's tool set is the intersection of the profile's tools and the parent's tools, and
run_codeis always removed. - Approval is always "never". A child cannot widen its own permissions; operations that need approval are rejected automatically.
- Cost is capped. Model, reasoning effort, tokens, and recursion depth are all bounded; out-of-range values fail loudly instead of silently downgrading.
Data
~/.dsh/subagent-profiles.json— the profile registry (edited from the settings page).~/.dsh/subagent-profiles.state.json— the plugin's enable/disable switch (default enabled).~/.dsh/subagent-profiles.failed-traces.json— the failure ledger (dispatch failure traces).~/.dsh/.agent-presets/orchestrator/— the self-installedorchestratoragent preset (synced from the bundledpresets/orchestrator/on every startup).
DSH_HOME is respected and defaults to ~/.dsh. Uninstalling the plugin removes the three data files above and the self-installed orchestrator preset directory (other plugins' presets are left untouched); re-installing or re-launching re-syncs the preset and regenerates the data files.
Known limitations
- Background one-shot dispatch requires
@deepseek-ai/dsh-jobsand@deepseek-ai/dsh-tool-jobsto be loaded; otherwise it fails with "dispatch: 后台派发不可用:缺少 jobs 服务". - Continuable mode goes through the DSH standard composition path, so the
presetswap andreasoningEffortare ignored (the child inherits the parent preset at the default reasoning effort). - Continuable tool gate is a plugin-side mitigation: the child's
allowis pre-computed as a closed set — parent tool set −run_code−deny, then intersected withallow. Assumption: continuable inherits the parent preset, so the child's tool set ≈ the parent's. Failure condition: any host behavior change that makes the child's tool set differ from the parent's (not only preset swap — e.g. a future preset swap, composing a different tool set) means the parent set can contain tools the child does not have, sotools.restrictthrows "unknown tool" and this mitigation automatically degrades to fail-loud (conservatively safe); it must then be replaced with a true parent ∩ child intersection once the upstream provides a provider guard seam.
Repository structure
dsh-subagent-profile/
├── index.mjs # host side: the plugin itself (dispatch tool, profile provider, service, HTTP routes)
├── lib/
│ ├── client.js # browser side: settings page + dispatch tool-call card
│ └── core/ # host-side modules (lightweight layering)
│ ├── pure.mjs # dependency-free pure functions (sanitize / prune / guard math — unit-tested)
│ ├── shims.mjs # the single @deepseek-ai import facade (guards fail loud, helpers degrade softly)
│ ├── catalog.mjs # tool-name → zh / category tables (zero-dep)
│ ├── catalog-cache.mjs # process-shared catalog snapshot (models / presets / tools, TTL cache)
│ ├── cost-guard.mjs # runtime capability checks (provider / model / reasoning effort)
│ ├── decision-trace.mjs # decision trace (gates / effective / settled) + failure ledger
│ ├── delegation.mjs # background one-shot settling + child usage collection
│ ├── dispatch-gates.mjs # pre-dispatch gates (whitelist / cost / intersection / budget)
│ ├── dispatch-guard.mjs # concurrency + per-parent token budget guard
│ ├── dispatch-schema.mjs # dispatch tool input/output schema declarations
│ ├── dispatch-tool.mjs # dispatch tool factory (defineTool + execute + syncTool)
│ ├── escape.mjs # escape-hatch allow store
│ ├── evolution-ledger.mjs # dispatch ledger (jsonl) + governance audit
│ ├── evolution-summary.mjs # T1 aggregate summaries + advice text
│ ├── http-routes.mjs # settings loopback HTTP routes
│ ├── intersection.mjs # tool-intersection pure core
│ ├── presets-sync.mjs # bundled preset self-install (hash-gated sync)
│ ├── profile-provider.mjs # the `profile` subagent provider
│ ├── profiles-store.mjs # profile registry store + switch persistence
│ └── whitelist.mjs # system-trust preset whitelist
├── presets/orchestrator/ # bundled "orchestrator" agent preset (self-installed, synced on every startup)
├── cordis.patch.yml # bundle patch: inserts the plugin row into the host composition
├── .gitea/workflows/ci.yml # bare-CI (Gitea Actions; needs an Act runner on the server)
├── package.json # metadata, files whitelist, exports (test / test:bare / preflight scripts)
├── scripts/
│ ├── preflight.mjs # preflight: preset-tree reconciliation + no hardcoded version badge (zero-dep)
│ └── leak-scan.mjs # public-release gate: scans all history + worktree for sensitive patterns
├── docs/
│ └── screenshots/ # README screenshots
├── test/ # host-side tests (node:test, zero extra deps; 352 cases — bare subset in CI, junction local)
│ ├── README.md / README.zh.md # test directory guide (EN/ZH) — two-tier split explained
│ ├── harness/ctx.mjs # fake Cordis ctx + ~/.dsh isolation
│ ├── pure / input-schema / catalog-integrity.test.mjs # bare tier (import-free, runs in bare CI)
│ └── *.test.mjs # junction tier (local only): characterization / facade / gating / persist / recycle /
│ # cost-guard / continuable-guard / decision-trace / dispatch-guard / escape-hatch /
│ # evolution-* / csrf / label-preset-sync / percall-spec / trust-label / audit-meta / …
├── README.md / README.zh.md # this document (EN/ZH)
└── LICENSE
Contributing
Found a bug or have an idea? Open an issue or submit a pull request — all contributions are welcome.
If this plugin has been useful to you, please give it a ⭐ on GitHub — it helps others find it.
Credits
The bundled orchestrator agent preset was inspired by dsh-liangshen (梁神模式) from dsh-web-ui, licensed under Apache-2.0. Thanks to its author for the great work.
License
MIT — Copyright (c) 2026 muzyLink
Install
Install the catalog once, then DeepSeek Harness can find and install any plugin from this site automatically:
dsh plugin add dshbase-catalog Then say "install dsh-subagent-profile for me" — your agent finds it in the directory and installs it. Docs: dshbase-catalog · verified packs.
This plugin is GitHub source (not published to npm) — install it straight from the repo:
Web profile:
dsh plugin --profile web add github:muzyLink/dsh-subagent-profile Headless (CLI) profile:
dsh plugin --profile headless add github:muzyLink/dsh-subagent-profile Test report
Verified: L1 install + L2 load + L3 runtime from GitHub source on dsh 0.1.0-rc.6.