Dudu-0223
0fadb08fbd
Proactive pressure scaled `thresholdRatio` by the routed model's entire
context window, but a request also reserves output tokens the provider
charges to that same window — the LLM service deliberately never clamps
them (`packages/llm/llm/src/index.ts`).
With a 1,000,000-token window and the default 256,000-token request cap the
gate therefore sat at 800,000 while the provider refused any prompt above
792,576. Observed in a real session: three requests died at 792,850 /
797,639 / 793,352 message tokens ("maximum context length is 1048576 ...
you requested 1048850 tokens"), no pressure compaction ever fired, and the
overflow recovery that did run could not condense an ~800k-token history
into the 8192-token summary cap.
Scale both ratios by the message budget left after the reserve, sourced from
the effective request envelope's `maxTokens` and falling back to the
adapter's per-request output default. A reserve that leaves no message
budget reports a target-specific pressure configuration error, matching the
existing per-route warning policy rather than silently disabling proactive
compaction. The headless compaction e2e advertises a coherent capacity/cap
pair so its 8000-token message budget and 4000-token threshold are unchanged.
Related #1841
@deepseek-ai/dsh
English | 中文
The dsh command is the sole supported Node application launcher: profiles are ordered stacks of plugin-bundle patch layers under the user's own overrides. SDK and ACP are profiles, not separate public bins. The Python runtime wheel packages this same command; the SDK defaults to sdk, and the minimal example selects sdk-minimal. src/args.ts owns the command grammar, and src/bin.ts loads only the selected runner. Invalid commands, options from another mode, and fatal configuration or boot failures exit nonzero.
Entry modes
| Command | Purpose |
|---|---|
dsh <name> / dsh --profile <name> |
Boot the named profile under $DSH_HOME/profiles/<name>. |
dsh --profile <name> --from-default-profile <template> |
Create a new custom profile from a shipped template, then boot it. |
dsh --profile acp |
Serve automation clients over ACP stdio until disconnect. |
dsh --profile headless "job" |
Run one fresh persisted session, print the final answer, and exit. |
dsh --profile sdk |
Serve SDK clients over JSON-RPC stdio until shutdown or disconnect. |
dsh --profile sdk-minimal |
Serve SDK clients with the standalone minimal agent tree. |
dsh web |
Boot the Web profile. |
dsh plugin --profile <name> <pnpm args> |
Manage a profile's plugins by forwarding to pnpm in the profile directory. |
The invoking directory is the default workspace root. The web, headless, sdk, sdk-minimal, and acp profiles auto-initialize on first use from shipped templates. Create another profile at an unused, non-shipped name with --from-default-profile, or initialize a base-backed profile through dsh plugin. The desktop name is reserved for the Electron-owned profile, so the CLI rejects boot, config-dump, and plugin-management requests for it.
App arguments
The launcher parses only its own flags and hands everything after them to the booted profile, where any injected app plugin may parse the shared immutable snapshot (dsh-cmdline). The first token the launcher does not recognize starts the app's arguments:
dsh --profile web --port 8080 # --port belongs to the web app
dsh --profile tui --resume <id> # example, assuming the tui profile is installed; --resume belongs to the terminal app
dsh --profile headless "run the tests"
dsh --profile web --help # the web app's flags, not the launcher's
dsh --help # the launcher's own help
Profiles
A profile directory holds a package.json (out-of-tree plugin dependencies plus the profile manifest dsh.profile with its ordered bundles list) and a cordis.patch.yml (the user's own patch layer). dsh-hmr, when enabled in YAML, watches the profile manifest and both profile and home patch files, then recomposes all layers through one serialized reload. Without HMR, changes apply on restart. Edits arriving during watcher registration use the same nonfatal reload reporting as later edits. Plugin Manager shares package operations and the profile write lock with dsh plugin; package updates retain disabled bundle selections. CLI package commands inherit authentication variables and terminal descriptors, including interactive build approval; service calls retain their scrubbed environment and captured diagnostics.
The tree composes over an empty root:
- each bundle's patch in
dsh.profile.bundlesorder - then the profile's
cordis.patch.yml, then the home-level$DSH_HOME/cordis.patch.yml - then
--patchoverlays
Bundles named in dsh.profile.bundles resolve from the dsh installation first (@deepseek-ai/dsh-base, @deepseek-ai/dsh-web-app, @deepseek-ai/dsh-headless, @deepseek-ai/dsh-sdk-app, @deepseek-ai/dsh-sdk-minimal, @deepseek-ai/dsh-acp-app), then from the profile's own node_modules, where pnpm installs out-of-tree plugins.
Use --dump-default-config and --dump-config to inspect the composed tree without booting it.
The CLI behavior reference owns exact layer precedence, flags, shutdown behavior, deployment defaults, and source execution. The startup and reload failure table compares optional and required plugin failures with configuration HMR.
Optional overlays
config/examples/ ships opt-in overlays for GitHub review webhooks, session-local Schedule, memory MCP servers, and runtime Cordis tools. They are never part of a default profile; the user guides and developer practice guides own setup and safety instructions.
Development
Production runs require built package and frontend artifacts. From the repository root, run pnpm run build separately, then use pnpm dsh <args...> to run the TypeScript entry and forward every argument; the source-execution reference owns the module-resolution contract.
The @deepseek-ai/dsh/profile-boot export provides the shared profile lifecycle to the Desktop host. A resolved application profile supplies its own installation anchor for runtime package resolution while retaining the Harness home patch, proxy environment, telemetry switch, patch reload, and bounded shutdown.
The Web failure matrix runs the built CLI through startup failures and native configuration HMR with awaitWriteFinish enabled in test:expected. It verifies authenticated HTTP responses, diagnostics, recovery, process exits, and disposal without model API calls; the startup acceptance also covers the shipped required Web dependencies and port conflicts.