@agi-fans/oh-my-dsh 0.17.0 → 0.19.0

package/README.md

@@ -10,12 +10,14 @@ omdsh is a focused, keyboard-first DeepSeek coding agent built on the plugin arc
 
 [![CI](https://github.com/agi-fans/oh-my-dsh/actions/workflows/ci.yml/badge.svg)](https://github.com/agi-fans/oh-my-dsh/actions/workflows/ci.yml) [![npm version](https://img.shields.io/npm/v/%40agi-fans%2Foh-my-dsh?style=flat-square&logo=npm)](https://www.npmjs.com/package/@agi-fans/oh-my-dsh) [![npm downloads](https://img.shields.io/npm/dm/%40agi-fans%2Foh-my-dsh?style=flat-square&logo=npm)](https://www.npmjs.com/package/@agi-fans/oh-my-dsh) [![Node.js ^22.19 or >=24](https://img.shields.io/badge/node-%5E22.19%20%7C%7C%20%3E%3D24-339933?style=flat-square&logo=node.js)](https://nodejs.org/) [![MIT License](https://img.shields.io/npm/l/%40agi-fans%2Foh-my-dsh?style=flat-square)](https://github.com/agi-fans/oh-my-dsh/blob/main/LICENSE)
 
-English · [简体中文](https://github.com/agi-fans/oh-my-dsh/blob/main/README.zh-CN.md)
+[Documentation](https://omdsh.agi.fans/) · English · [简体中文](https://github.com/agi-fans/oh-my-dsh/blob/main/README.zh-CN.md)
 
 </div>
 
 ![oh-my-dsh terminal interface](https://raw.githubusercontent.com/agi-fans/oh-my-dsh/main/apps/site/public/screenshot.webp)
 
+Screenshot from v0.5.1. See [Keyboard and keys](https://omdsh.agi.fans/docs/keyboard/) for the current turn presentation and navigation.
+
 ## Quick start
 
 Requirements: Node.js 22.19 or later in the 22.x line, or Node.js 24 or newer, plus a DeepSeek API key for live model turns.
@@ -32,20 +34,45 @@ Run `/login` once inside omdsh to validate and save your DeepSeek API key, then
 - **Durable conversations:** search, pin, rename, and resume sessions in `/sessions`; rewind, retry, compact, and export complete transcripts as Markdown or standalone HTML.
 - **Three real session controls:** choose a Harness Agent preset (Standard, PTC, Minimal, or Cordis), Workflow (Default or Plan), and Access (Read only, Workspace write, or Full access). Each Agent preset owns its tool exposure; PTC uses the generated TypeScript SDK automatically.
 - **Rich terminal input:** mention project files and other sessions with `@`, paste clipboard images, reuse persistent prompt history, edit multiline prompts externally, and retrieve queued follow-ups.
-- **Readable tool activity:** follow streaming calls and live subagent progress, press Down on an empty composer then Enter (or use Alt+A directly) to select a child in the keyboard-driven Agent Hub, steer a continuable child from its transcript, inspect distinct Input and Output sections, expand long results, and keep domain-specific presentation owned by tool plugins.
+- **Readable turns:** follow thinking, replies, and tool previews while the agent works. Completed work folds to a duration summary and the full final answer; `Ctrl+O` restores the process view. Scroll through either view, keep the current prompt pinned, and jump back to the latest message with `End`.
+- **Inspectable work:** press `Alt+O` for full tool inputs and results in the current turn, use `/trajectory` for event details, see the files changed by each turn, and let the model search its earlier sessions in the same workspace. Open the Agent Hub with `Alt+A` to follow child agents and steer continuable tasks.
 - **Live operational context:** see Agent, Workflow, Access, model, reasoning effort, workspace, Git state, context pressure, tokens, TTFT, throughput, cache, timings, turns, and steps without leaving the composer; use `/context` for an inline projection-backed breakdown that remains in the transcript.
 - **Responsive by design:** retain settled transcript layout, coalesce scroll updates, emit row-level terminal diffs, and preserve correct display-cell alignment for CJK text and emoji.
 
-## Learn
+## Documentation
+
+The full documentation is published at [omdsh.agi.fans](https://omdsh.agi.fans/), and the Markdown under `apps/site/content/` is its source of truth.
+
+**Start here**
+
+- [Tutorials](https://omdsh.agi.fans/docs/tutorials/) — complete a first task, give the agent precise context, guide queued work, recover long sessions, customize the environment, and write an installable plugin.
+- [Commands](https://omdsh.agi.fans/docs/commands/) — every slash command with its arguments and aliases.
+- [Keyboard and keys](https://omdsh.agi.fans/docs/keyboard/) — editing, transcript, overlay keys, and `keybindings.json`.
+
+**Reference**
 
-The full documentation is also published on the [official site](https://omdsh.agi.fans/).
+- [Sessions and history](https://omdsh.agi.fans/docs/sessions/) — durable logs, the Session Library, and the local data files.
+- [Subagents and delegation](https://omdsh.agi.fans/docs/subagents/) — the three transports, the Agent Hub, steering, and background jobs.
+- [Observability](https://omdsh.agi.fans/docs/observability/) — `/trajectory`, `/context`, `/diff`, `/tools`, and `/mcp`.
+- [Settings](https://omdsh.agi.fans/docs/settings/) — appearance, motion, notifications, and the configurable status line.
+- [Permissions and access](https://omdsh.agi.fans/docs/permissions/) — Access presets, sandbox, and approvals.
+- [Command line](https://omdsh.agi.fans/docs/cli/) — flags, subcommands, and environment variables.
+- [Tools](https://omdsh.agi.fans/docs/tools/) — what the model can use, grouped by capability.
+- [Troubleshooting](https://omdsh.agi.fans/docs/troubleshooting/) — exit, color, scrollback, and boot failures.
 
-- [Tutorials](https://omdsh.agi.fans/docs/tutorials/) — complete a first task, add precise context, guide queued work, recover long sessions, customize the environment, and write an installable plugin.
-- [Skills and MCP](https://omdsh.agi.fans/docs/skills-and-mcp/) — extend a project with reusable instructions and external tools.
+**Extend and integrate**
+
+- [Skills and MCP](https://omdsh.agi.fans/docs/skills-and-mcp/) — reusable instructions and external tools.
+- [Language servers](https://omdsh.agi.fans/docs/language-servers/) — configure read-only code intelligence.
 - [User plugins](https://omdsh.agi.fans/docs/plugins/) — install DSH bundles into the omdsh Profile with `omdsh plugin`.
-- [Architecture](https://omdsh.agi.fans/docs/architecture/) — understand the plugin boundaries and runtime data flow.
-- [Performance](https://omdsh.agi.fans/docs/performance/) — inspect the benchmarks, methodology, and rendering optimizations.
-- [Report a bug or request a feature](https://github.com/agi-fans/oh-my-dsh/issues/new/choose) — use a guided form to include the version, environment, reproduction steps, and sanitized context.
+- [Plugin internals](https://omdsh.agi.fans/docs/plugin-internals/) — the `ctx.tui` surface and the contribution layer.
+
+**Internals**
+
+- [Architecture](https://omdsh.agi.fans/docs/architecture/) — plugin boundaries and runtime data flow.
+- [Performance](https://omdsh.agi.fans/docs/performance/) — benchmarks, methodology, and rendering optimizations.
+
+[Report a bug or request a feature](https://github.com/agi-fans/oh-my-dsh/issues/new/choose) with a guided form for the version, environment, reproduction steps, and sanitized context.
 
 ## Why oh-my-dsh
 
@@ -78,7 +105,7 @@ The TUI package is split into a service definition, local terminal Provider, ses
 
 ## Performance
 
-Performance is part of the TUI architecture: durable sessions replay in linear time, Harness Projections avoid repeated history scans, settled transcript blocks retain formatted layout, and the terminal writer emits row-level diffs. On the documented Apple M5 Pro environment, restoring 10,000 conversation turns takes a median 2.62 ms, 10,000 tool calls take 22.71 ms, and rendering cached frames over a 5,000-turn surface averages 0.35 ms per frame.
+Performance is part of the TUI architecture: durable sessions replay in linear time, Harness Projections avoid repeated history scans, settled transcript blocks retain formatted layout, and the terminal writer emits row-level diffs. The documented Apple M5 Pro measurements from August 2026 recorded a median 2.62 ms for restoring 10,000 conversation turns, 22.71 ms for 10,000 tool calls, and 0.35 ms per cached frame over a 5,000-turn surface. These are historical measurements, not a benchmark of the latest version.
 
 See the reproducible [TUI performance report](https://omdsh.agi.fans/docs/performance/) or run `pnpm benchmark:tui` locally.
 
@@ -86,9 +113,9 @@ See the reproducible [TUI performance report](https://omdsh.agi.fans/docs/perfor
 
 Run `/login` to configure a provider. DeepSeek still opens the official key dashboard, validates the key, and prefers the stored credential over an inherited `DEEPSEEK_API_KEY`. When a mounted provider registers a Harness authorization flow, `/login` lists that flow and its methods and the terminal only renders the notices and prompts the flow asks for. The same command can also activate a catalog API-key provider or add a custom provider with its own id, base URL, protocol, and model ids. `/model` then lists every live route. `/logout` removes an omdsh-managed choice and, for DeepSeek, falls back to the environment when available.
 
-Model settings can also come from `$DSH_HOME/settings.yaml`. Use `/model favorite` and `/model unfavorite` to maintain a small local cycle list, then use `Ctrl+P`/`Alt+P` to move through it and `Ctrl+T` to change reasoning effort. Skills and MCP configuration are documented in [Skills and MCP](https://omdsh.agi.fans/docs/skills-and-mcp/).
+Model settings and the rest of the plugin config persist into the active Profile's Cordis patch, `$OMDSH_HOME/profiles/omdsh/cordis.patch.yml`; a `settings.yaml` left by an earlier release is imported once at startup and renamed to `settings.yaml.imported`. Use `/model favorite` and `/model unfavorite` to maintain a small local cycle list, then use `Ctrl+P`/`Alt+P` to move through it and `Ctrl+T` to change reasoning effort. Skills and MCP configuration are documented in [Skills and MCP](https://omdsh.agi.fans/docs/skills-and-mcp/).
 
-After an upgrade, omdsh can show release notes once at startup. Use `/changelog` for recent entries or `/changelog full` for the complete packaged history. A cached daily npm check reports newer versions without installing anything automatically; both behaviors can be customized in `/settings`. Terminal notifications for completed long turns and human prompts are also configured there and remain off by default.
+After an upgrade, omdsh can show release notes once at startup. Use `/changelog` for recent entries or `/changelog full` for the complete packaged history. A cached daily npm check reports newer versions without installing anything automatically; both behaviors can be customized in `/settings`. Terminal notifications for completed long turns and human prompts are also configured there and remain off by default. Every setting, its allowed values, and its default are listed in [Settings](https://omdsh.agi.fans/docs/settings/).
 
 Generate shell integration without starting the TUI or making network requests:
 

package/config/cordis.yml

@@ -13,8 +13,13 @@
   - id: llm
     name: '@deepseek-ai/dsh-llm'
 
+  # `dsh-llm-deepseek` is the Messages transport library since the 0.2.0
+  # corridor; the plugin that owns the `deepseek-official` route now lives in
+  # `dsh-llm-deepseek-api-key` (the account-token sibling owns `deepseek-account`).
+  # The row id stays `llm-deepseek` because the settings namespace the TUI
+  # `/auth` and `/model` commands read is derived from the entry id.
   - id: llm-deepseek
-    name: '@deepseek-ai/dsh-llm-deepseek'
+    name: '@deepseek-ai/dsh-llm-deepseek-api-key'
     config:
       # Current built-ins share this fallback. Exact catalog values still win,
       # and omitting models keeps future adapter additions visible.
@@ -63,9 +68,25 @@
   - id: session-query
     name: '@deepseek-ai/dsh-session-query-sqlite'
     config:
-      path: ':memory:'
+      # Durable derived index rather than a per-process one: a cold launch then
+      # reconciles only new or changed logs instead of re-indexing the whole
+      # store, which is what made the first `/resume <query>` cost ~14s.
+      path: !!js "(process.env.OMDSH_HOME || process.env.DSH_HOME || process.env.HOME + '/.dsh') + '/sessions-query.sqlite'"
       openAt: first-search
 
+  # Model-facing session history: five read-only tools that let the agent find
+  # work it did in an earlier session. Without them a model that solved
+  # something three sessions ago can only be told again, because `/sessions`
+  # and `/resume` are human affordances it cannot reach. Cross-session reads
+  # are authorized only when the target session's cwd exactly matches the
+  # caller's, so the model cannot read another workspace's history.
+  #
+  # The cost is five tool schemas plus fixed guidance on every request. A
+  # deployment that would rather not pay it turns the row off in a home
+  # overlay with `disabled: true`.
+  - id: tool-session-query
+    name: '@deepseek-ai/dsh-tool-session-query'
+
   - id: session-reference
     name: '@deepseek-ai/dsh-session-reference'
 
@@ -99,6 +120,14 @@
   - id: workspace
     name: '@deepseek-ai/dsh-workspace'
 
+  # Per-turn changed-file summary. The plugin diffs git working-tree snapshots
+  # taken at turn start and turn end and announces each result with a
+  # `workspace/changes` event; the transcript renders the served file list, and
+  # a resumed session falls back to the turn-only form because the summaries
+  # live on the host only while the Session does.
+  - id: workspace-changes
+    name: '@deepseek-ai/dsh-workspace-changes'
+
   # Workspace @file discovery. The TUI owns mention syntax; this provider
   # ranks path-only candidates and, when `read` is visible, adds the stable
   # file-reference instruction. Selecting a row still inserts a path.
@@ -110,14 +139,18 @@
     config:
       dshHome: !!js process.env.OMDSH_HOME || process.env.DSH_HOME || undefined
 
-  # User settings document ($DSH_HOME/settings.yaml) and credential sources
-  # (inherited env, managed .credentials.yaml, project/user .env) — the same
-  # rows the harness base mounts, so omdsh honors the user's existing model
-  # route and API key setup out of the box.
+  # User settings and credential sources (inherited env, managed
+  # .credentials.yaml, project/user .env) — the same rows the harness base
+  # mounts, so omdsh honors the user's existing model route and API key setup
+  # out of the box. Settings are live plugin config: forms expose the volatile
+  # fields of active entries, and config-editor persists an edit into the
+  # profile patch. A settings.yaml left by an earlier release is imported once
+  # and renamed to settings.yaml.imported.
+  - id: config-editor
+    name: '@deepseek-ai/dsh-config-editor'
+
   - id: settings
-    name: '@deepseek-ai/dsh-settings-file'
-    config:
-      dshHome: !!js process.env.OMDSH_HOME || process.env.DSH_HOME || undefined
+    name: '@deepseek-ai/dsh-settings'
 
   - id: credentials
     name: '@deepseek-ai/dsh-credentials-local'
@@ -146,18 +179,24 @@
       provider: !!js process.env.OMDSH_PROVIDER || 'deepseek-official'
       model: !!js process.env.OMDSH_MODEL || 'deepseek-flash'
 
-  - id: agent-presets
-    name: '@deepseek-ai/dsh-agent-presets'
+  # Agent presets: the registry owns the roster and the default, and every
+  # preset is an ordinary `@deepseek-ai/dsh-agent-preset` row declaring its
+  # child plugins. omdsh ships `standard`, `code`, `minimal`, and `cordis` as
+  # insert patches under `config/presets/`, applied after this file.
+  - id: agent-preset-registry
+    name: '@deepseek-ai/dsh-agent-preset-registry'
     config:
       default: standard
-      includeShippedRoot: false
-      roots:
-        - path: !!js "process.getBuiltinModule('node:url').fileURLToPath(new URL('./agent-presets/', baseUrl))"
-          trust: system
 
   - id: commands
     name: '@deepseek-ai/dsh-commands'
 
+  # `/feedback <text>` records one log-only `feedback/record` event: immediate,
+  # model-invisible, and no report leaves the machine. The Web feedback dialog
+  # behind the same service stays a host-plane surface omdsh does not mount.
+  - id: command-feedback
+    name: '@deepseek-ai/dsh-command-feedback'
+
   - id: user-questions
     name: '@deepseek-ai/dsh-user-questions'
 
@@ -322,8 +361,8 @@
     config:
       agents: []
 
-  - id: code-runtime
-    name: '@deepseek-ai/dsh-code-runtime-worker-thread'
+  - id: ptc-runtime
+    name: '@deepseek-ai/dsh-ptc-runtime-node'
 
   - id: cordis-host-runner
     name: '@deepseek-ai/dsh-cordis-host-runner'
@@ -335,6 +374,8 @@
 
   - id: tool-ask-user
     name: '@deepseek-ai/dsh-tool-ask-user'
+    config:
+      mode: legacy
 
   - id: plan-mode
     name: '@deepseek-ai/dsh-plan-mode'
@@ -377,34 +418,52 @@
   - id: tool-present
     name: '@deepseek-ai/dsh-tool-present'
 
-  # Anonymous public web fetch, no credential and no search provider: a fetch
-  # costs nothing and the TUI already renders web sources. Search stays off
-  # because DeepSeek's native search spends a whole model turn per query; a
-  # deployment that wants it mounts a provider in its own overlay and flips
-  # `search` on here.
+  # Web tools: the model can fetch public pages for free, and search the web
+  # through DeepSeek's hosted search. Search is on by default because the
+  # official provider ships with the base and speaks the same Messages protocol
+  # as the conversation route, so it needs no extra credential setup: the key
+  # resolves through the credentials service at each search. A search is a full
+  # auxiliary model request, so its cost is bounded by `maxUses` (and the
+  # provider's own settings section tunes model, token cap, and endpoint); a
+  # deployment that wants no search disables `tool-web`'s `search` in its own
+  # overlay.
   - id: web
     name: '@deepseek-ai/dsh-web'
 
+  - id: web-search-deepseek
+    name: '@deepseek-ai/dsh-web-search-deepseek'
+    config:
+      apiKeyEnv: DEEPSEEK_API_KEY
+      maxUses: 5
+
   - id: web-fetch-http
     name: '@deepseek-ai/dsh-web-fetch-http'
 
   - id: tool-web
     name: '@deepseek-ai/dsh-tool-web'
     config:
-      search: false
+      search: true
       fetch: true
-
-  # Spills one oversized plain-text tool result to a private file and hands the
-  # model a bounded preview plus the path. The cap is deliberately far above
-  # upstream's 50 KB: the session log carries the replacement, so a lower cap
-  # would shorten what a human can expand in the transcript.
+      searchTimeoutMs: 60000
+
+  # MCP resources: the three shared list/read tools appear only for callers whose
+  # scope has a connected MCP server, so a profile with no `mcp.json` pays no
+  # catalog cost. Resource text enters the conversation; binary payloads stay
+  # programmatic.
+  - id: mcp-resources
+    name: '@deepseek-ai/dsh-mcp-resources'
+
+  # Spills one oversized tool result to a private file and hands the model a
+  # bounded preview plus the path. The cap is deliberately far above upstream's
+  # 12500 estimated tokens: the session log carries the replacement, so a lower
+  # cap would shorten what a human can expand in the transcript.
   - id: spill-local
     name: '@deepseek-ai/dsh-spill-local'
 
   - id: spill-policy
     name: '@deepseek-ai/dsh-spill-policy'
     config:
-      maxInlineBytes: 200000
+      maxInlineTokens: 50000
 
   - id: token-meter
     name: '@deepseek-ai/dsh-token-meter'
@@ -419,6 +478,12 @@
       headChars: 4096
       tailChars: 1024
 
+  # When an image-capable route rejects a request as over its image budget,
+  # replace the oldest images with placeholders on the surface and retry. The
+  # route keeps its full history in the session log; only the request degrades.
+  - id: image-offload
+    name: '@deepseek-ai/dsh-compaction-image-offload'
+
   - id: compaction
     name: '@deepseek-ai/dsh-compaction-basic'
 
@@ -524,13 +589,13 @@
 
   # Workflow and Ralph, as the upstream base bundle mounts them.
   #
-  # The worker-thread engine is the seam provider: `dsh-workflow-worker-thread`
-  # executes model-written orchestration scripts off the host event loop, which
+  # The PTC engine is the seam provider: `dsh-workflow-ptc` executes
+  # model-written orchestration scripts through the session's PTC runtime, which
   # is the same isolation the ACP subagent transport buys for delegation.
   # Mounting `dsh-workflow` as well would double-register `workflowEngine`, so
   # it stays a dependency the engine resolves rather than a composition row.
-  - id: workflow-worker-thread
-    name: '@deepseek-ai/dsh-workflow-worker-thread'
+  - id: workflow-ptc
+    name: '@deepseek-ai/dsh-workflow-ptc'
     config:
       provider: spawn
 
@@ -545,9 +610,15 @@
 
   # Ralph iterates fresh agents over a build-time-fixed script; it reuses the
   # continuable in-process `spawn` provider, so it depends on that transport
-  # staying mounted.
+  # staying mounted. Off by default, matching the upstream base bundle: the
+  # tool description restricts `ralph` to runs the human explicitly asked for,
+  # and completion is a worker self-report rather than an independent
+  # evaluation. A home overlay restores it:
+  #   - id: tool-ralph
+  #     disabled: false
   - id: tool-ralph
     name: '@deepseek-ai/dsh-tool-ralph'
+    disabled: true
     config:
       subagentProvider: spawn
       maxRounds: 64
@@ -556,6 +627,10 @@
     name: '@agi-fans/dsh-tui'
     config:
       model: !!js process.env.OMDSH_MODEL || 'deepseek-flash'
+      # The settings overlay edits the Profile patch for optional features;
+      # `disabled` is a Loader option the settings service cannot write, so
+      # this row carries the home whose patch the Features section owns.
+      dshHome: !!js process.env.OMDSH_HOME || process.env.DSH_HOME || process.env.HOME + '/.dsh'
 
   - id: agent-behavior
     name: '@agi-fans/oh-my-dsh/agent-behavior'
@@ -565,6 +640,11 @@
 
   - id: tui-session-runtime
     name: '@agi-fans/dsh-tui/session-runtime'
+    config:
+      # Product state this runtime owns: Session Library rows reused across
+      # launches (without them a cold process reads every stored log again) and
+      # the marker that upgrades stored sessions once per session format.
+      stateDir: !!js "(process.env.OMDSH_HOME || process.env.DSH_HOME || process.env.HOME + '/.dsh') + '/omdsh'"
 
   - id: tui-human-interaction
     name: '@agi-fans/dsh-tui/human-interaction'

package/config/presets/code.patch.yml

@@ -0,0 +1,40 @@
+# Agent preset code: one `@deepseek-ai/dsh-agent-preset` declaration.
+# Edits saved by `omdsh plugin` or a profile patch override this row's
+# `config.plugins` by id from the profile patch.
+- insert:
+    - id: preset-code
+      name: '@deepseek-ai/dsh-agent-preset'
+      config:
+        id: code
+        name: PTC
+        description: Standard coding agent optimized for programmatic tool calling through the generated TypeScript SDK.
+        order: 2
+        plugins:
+          - id: persona
+            name: '@deepseek-ai/dsh-persona'
+            config:
+              prefix: >-
+                You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}. Prefer programmatic tool calling for multi-step operations when the generated tools SDK is available.
+
+          # PTC presents the tool registry as a generated SDK over `run_code`, so the
+          # general-purpose `workflow_run` tool would be a second model-authored
+          # orchestration surface beside it. The upstream `ptc` preset disables its own
+          # workflow rows for that reason; a preset cannot disable a host row, so the
+          # restriction does it here. `ralph` needs no entry: the deployment ships it
+          # disabled, and `tools.restrict()` rejects a name that no global tool owns — a
+          # home overlay that restores it accepts a second surface in this preset.
+          - id: profile
+            name: '@agi-fans/dsh-tui/agent-profile'
+            config:
+              tools:
+                deny:
+                  - workflow_run
+
+          # PTC for this preset alone; `standard`, `minimal`, and `cordis` inherit the
+          # deployment default (`native`) from the host `tools` row. The row fails the
+          # mount when no PTC runtime is composed, which is the point: the session does
+          # not choose its presentation at runtime any more.
+          - id: tool-presentation
+            name: '@deepseek-ai/dsh-agent-tool-presentation'
+            config:
+              mode: ptc

package/config/presets/cordis.patch.yml

@@ -0,0 +1,22 @@
+# Agent preset cordis: one `@deepseek-ai/dsh-agent-preset` declaration.
+# Edits saved by `omdsh plugin` or a profile patch override this row's
+# `config.plugins` by id from the profile patch.
+- insert:
+    - id: preset-cordis
+      name: '@deepseek-ai/dsh-agent-preset'
+      config:
+        id: cordis
+        name: Cordis
+        description: Standard coding agent with read-only runtime inspection for authoring Harness compositions.
+        order: 4
+        plugins:
+          - id: persona
+            name: '@deepseek-ai/dsh-persona'
+            config:
+              prefix: |-
+                You are a coding agent powered by the {{model}} model, running on the DeepSeek Harness. Your working directory is {{cwd}}.
+
+                You can inspect the live Cordis runtime while authoring compositions and plugins. Host composition owns shared registries and backends; an Agent preset owns one session's tools, persona, prompt sections, and isolated services. Use cordis_inspect_list to discover providers, then cordis_inspect_query to read exact Service, Event, and Tool APIs. These tools are read-only; changing the runtime means editing composition files, not calling a tool.
+
+          - id: tool-cordis
+            name: '@deepseek-ai/dsh-tool-cordis'

package/config/presets/minimal.patch.yml

@@ -0,0 +1,64 @@
+# Agent preset minimal: one `@deepseek-ai/dsh-agent-preset` declaration.
+# Edits saved by `omdsh plugin` or a profile patch override this row's
+# `config.plugins` by id from the profile patch.
+- insert:
+    - id: preset-minimal
+      name: '@deepseek-ai/dsh-agent-preset'
+      config:
+        id: minimal
+        name: Minimal
+        description: Focused two-tool coding agent with a persistent Bash session and str_replace_editor.
+        order: 3
+        plugins:
+          - id: persona
+            name: '@deepseek-ai/dsh-persona'
+            config:
+              prefix: You are a helpful software engineer assistant.{{omdsh_agent_behavior}}
+              complete: true
+              includeRuntimeContext: false
+
+          # `shell` is an omdsh alias for the host's one shell tool (`bash` on POSIX,
+          # `pwsh` on Windows), so this two-tool shape holds on every platform without a
+          # name that `tools.restrict()` would reject.
+          - id: profile
+            name: '@agi-fans/dsh-tui/agent-profile'
+            config:
+              tools:
+                allow:
+                  - shell
+                  - str_replace_editor
+
+          - id: persistent-shell
+            name: cordis:group
+            group: true
+            isolate:
+              terminals: true
+            config:
+              - id: terminal
+                name: '@deepseek-ai/dsh-terminal'
+
+              - id: terminal-bash
+                name: '@deepseek-ai/dsh-terminal-bash'
+                disabled: !!js process.platform === 'win32'
+                config:
+                  timeoutMs: 300000
+
+              - id: persistent-bash
+                name: '@deepseek-ai/dsh-tool-bash-persistent'
+                disabled: !!js process.platform === 'win32'
+                config:
+                  timeoutMs: 300000
+
+              # The PTY backend is shared; the dialect selects PowerShell on Windows.
+              - id: terminal-pwsh
+                name: '@deepseek-ai/dsh-terminal-bash'
+                disabled: !!js process.platform !== 'win32'
+                config:
+                  shellDialect: pwsh
+                  timeoutMs: 300000
+
+              - id: persistent-pwsh
+                name: '@deepseek-ai/dsh-tool-pwsh-persistent'
+                disabled: !!js process.platform !== 'win32'
+                config:
+                  timeoutMs: 300000

package/config/presets/standard.patch.yml

@@ -0,0 +1,17 @@
+# Agent preset standard: one `@deepseek-ai/dsh-agent-preset` declaration.
+# Edits saved by `omdsh plugin` or a profile patch override this row's
+# `config.plugins` by id from the profile patch.
+- insert:
+    - id: preset-standard
+      name: '@deepseek-ai/dsh-agent-preset'
+      config:
+        id: standard
+        name: Standard
+        description: Full coding agent with files, shell, search, skills, planning, goals, and delegation.
+        order: 1
+        plugins:
+          - id: persona
+            name: '@deepseek-ai/dsh-persona'
+            config:
+              prefix: >-
+                You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}.

package/lib/agent-behavior.d.ts

@@ -2,23 +2,28 @@
  * Product-owned Agent behavior settings projected into the Harness system prompt.
  * @module @agi-fans/oh-my-dsh/agent-behavior
  */
-import type { Context } from '@deepseek-ai/cordis';
+import type { Context, Volatile } from '@deepseek-ai/cordis';
 import type { Agent } from '@deepseek-ai/dsh-agent';
 import type { AssembleContext } from '@deepseek-ai/dsh-system-prompt';
 import z from '@deepseek-ai/schemastery';
-import type { TuiAgentBehaviorSettings } from '@agi-fans/dsh-tui';
 export declare const name = "omdsh-agent-behavior";
 export declare const inject: string[];
-export declare const AGENT_BEHAVIOR_SETTINGS_NAMESPACE = "omdsh-agent";
+/** Profile entry id whose volatile config carries the language preference. */
+export declare const AGENT_BEHAVIOR_ENTRY = "agent-behavior";
 export declare const AGENT_BEHAVIOR_SECTION = "omdsh:agent-behavior";
 export declare const AGENT_BEHAVIOR_ORDER = 30;
 export declare const AGENT_BEHAVIOR_VARIABLE = "omdsh_agent_behavior";
 export declare const AGENT_LANGUAGES: readonly ["auto", "zh-CN", "en"];
 export type AgentLanguage = typeof AGENT_LANGUAGES[number];
-export interface AgentBehaviorSettings extends TuiAgentBehaviorSettings {
-    language: AgentLanguage;
+/** Live language setting; a settings form edit commits into this reference. */
+export interface Config {
+    language: Volatile<AgentLanguage>;
 }
-export declare const AgentBehaviorSettingsSchema: z<AgentBehaviorSettings>;
+export declare const Config: z<Schemastery.ObjectS<NoInfer<{
+    language: z<"auto" | "zh-CN" | "en", "auto" | "zh-CN" | "en", "volatile-defined">;
+}>>, Schemastery.ObjectT<NoInfer<{
+    language: z<"auto" | "zh-CN" | "en", "auto" | "zh-CN" | "en", "volatile-defined">;
+}>>, "plain">;
 /** Stable model instruction for one accepted language setting. */
 export declare function agentLanguageFragment(language: AgentLanguage): string;
 /** Freeze live user settings at a turn boundary while keeping direct assemblies useful. */
@@ -29,4 +34,4 @@ export declare class AgentBehaviorPrompt {
     fragment(context: AssembleContext, current: AgentLanguage): string;
     variable(context: AssembleContext, current: AgentLanguage): string;
 }
-export declare function apply(ctx: Context): void;
+export declare function apply(ctx: Context, config: Config): void;

package/lib/agent-behavior.js

@@ -5,13 +5,14 @@
 import z from '@deepseek-ai/schemastery';
 export const name = 'omdsh-agent-behavior';
 export const inject = ['settings', 'systemPrompt'];
-export const AGENT_BEHAVIOR_SETTINGS_NAMESPACE = 'omdsh-agent';
+/** Profile entry id whose volatile config carries the language preference. */
+export const AGENT_BEHAVIOR_ENTRY = 'agent-behavior';
 export const AGENT_BEHAVIOR_SECTION = 'omdsh:agent-behavior';
 export const AGENT_BEHAVIOR_ORDER = 30;
 export const AGENT_BEHAVIOR_VARIABLE = 'omdsh_agent_behavior';
 export const AGENT_LANGUAGES = ['auto', 'zh-CN', 'en'];
-export const AgentBehaviorSettingsSchema = z.object({
-    language: z.union([...AGENT_LANGUAGES]).default('auto'),
+export const Config = z.object({
+    language: z.union([...AGENT_LANGUAGES]).default('auto').volatile(),
 });
 const LANGUAGE_FRAGMENTS = {
     'zh-CN': '主要使用简体中文进行推理和面向用户的沟通，包括回复、提问、计划、待办文字和子 Agent 简报。代码、标识符、命令、Tool 参数、日志、引用、文件内容和惯用技术术语在准确性需要时保留原文或标准写法。项目指令中明确规定语言时，遵循项目指令；若用户明确要求当前任务使用其他语言，以该要求为准。',
@@ -44,11 +45,10 @@ export class AgentBehaviorPrompt {
         return fragment === '' ? '' : `\n\n${fragment}`;
     }
 }
-export function apply(ctx) {
-    const scope = ctx.settings.register(AGENT_BEHAVIOR_SETTINGS_NAMESPACE, AgentBehaviorSettingsSchema, { applies: 'live' });
+export function apply(ctx, config) {
     const prompt = new AgentBehaviorPrompt();
     ctx.on('agent/pre-step', async (payload, next) => {
-        prompt.snapshot(payload.agent, payload.turn, scope.get().language);
+        prompt.snapshot(payload.agent, payload.turn, config.language.get());
         return next();
     });
     ctx.on('agent/status', ({ agent, status }) => {
@@ -59,14 +59,15 @@ export function apply(ctx) {
     ctx.effect(() => ctx.systemPrompt.section({
         name: AGENT_BEHAVIOR_SECTION,
         order: AGENT_BEHAVIOR_ORDER,
-        text: context => prompt.fragment(context, scope.get().language),
+        text: context => prompt.fragment(context, config.language.get()),
     }), 'omdsh Agent behavior prompt section');
-    ctx.effect(() => ctx.systemPrompt.variable(AGENT_BEHAVIOR_VARIABLE, context => prompt.variable(context, scope.get().language)), 'omdsh Agent behavior prompt variable');
+    ctx.effect(() => ctx.systemPrompt.variable(AGENT_BEHAVIOR_VARIABLE, context => prompt.variable(context, config.language.get())), 'omdsh Agent behavior prompt variable');
     ctx.inject(['tui'], (tuiCtx) => {
         const binding = {
-            get: () => scope.get(),
-            update: next => scope.update({ language: next.language }),
-            watch: listener => scope.watch(next => { listener(next); }),
+            get: () => ({ language: config.language.get() }),
+            update: next => tuiCtx.settings.mutate(AGENT_BEHAVIOR_ENTRY, [{ op: 'set', path: ['language'], value: next.language }]),
+            // A form edit commits into the reference and reaches this fiber only.
+            watch: listener => ctx.on('loader/volatile-update', () => { listener({ language: config.language.get() }); }),
         };
         tuiCtx.effect(() => tuiCtx.tui.bindAgentBehaviorSettings?.(binding) ?? (() => { }), 'bind Agent behavior settings to TUI');
     });

package/lib/boot.d.ts

@@ -1,7 +1,7 @@
 /**
- * omdsh tree boot: mounts the omdsh Profile over an empty root, providing
- * the command line, the exit request, and the launch-environment snapshot
- * before any entry mounts.
+ * omdsh tree boot: mounts the omdsh Profile over an empty root, providing the
+ * command line, the profile facts, the runtime package resolution, and the
+ * launch-environment snapshot before any entry mounts.
  * @module @agi-fans/oh-my-dsh
  */
 import type { Context } from '@deepseek-ai/cordis';

package/lib/boot.js

@@ -1,14 +1,14 @@
 /**
- * omdsh tree boot: mounts the omdsh Profile over an empty root, providing
- * the command line, the exit request, and the launch-environment snapshot
- * before any entry mounts.
+ * omdsh tree boot: mounts the omdsh Profile over an empty root, providing the
+ * command line, the profile facts, the runtime package resolution, and the
+ * launch-environment snapshot before any entry mounts.
  * @module @agi-fans/oh-my-dsh
  */
-import { boot, healProfilesModuleFallback, installFailLoud } from '@deepseek-ai/dsh-app-boot';
+import { boot, createRuntimeResolution, installFailLoud, PluginPackages, } from '@deepseek-ai/dsh-app-boot';
 import { provideCmdline } from '@deepseek-ai/dsh-cmdline';
 import { DSH_LAUNCH_ENVIRONMENT_KEY } from '@deepseek-ai/dsh-launch-environment';
 import { NAME, prepareLaunchEnvironment } from "./composition.js";
-import { composeLaunch, INSTALL_ANCHOR } from "./profile.js";
+import { composeLaunch, INSTALL_ANCHOR, PROFILE_NAME } from "./profile.js";
 import { omdshHome } from "./config-paths.js";
 import { createProcessShutdown } from "./process-shutdown.js";
 export { NAME } from "./composition.js";
@@ -31,14 +31,37 @@ export async function runOmdsh(prompt, resume) {
     installFailLoud(NAME, process, async () => { await app.current?.fiber.dispose(); });
     const environment = prepareLaunchEnvironment();
     const composed = composeLaunch();
-    await healProfilesModuleFallback({
+    const home = omdshHome();
+    // The Profile facts every profile-scoped plugin reads: `dsh-settings` and
+    // `config-editor` persist edits into `patchPath`, and the composition
+    // preflight judges profile rows against the running DSH release. Overlays
+    // are the layers above the Profile patch, so a reconciliation re-derives
+    // exactly the list `composeLaunch` mounted.
+    const profileContext = {
+        name: PROFILE_NAME,
+        dir: composed.profile.dir,
+        patchPath: composed.profile.patchPath,
+        installAnchor: INSTALL_ANCHOR,
+        startedBundles: composed.profile.layers.map(layer => layer.packageName),
+        cwd: process.cwd(),
+        home,
+        overlays: composed.overlays,
+        telemetryDisabledEnv: process.env.DSH_TELEMETRY_DISABLED,
+    };
+    // One immutable package table for this launch: installation packages first
+    // (every `@deepseek-ai/*` and `@agi-fans/dsh-tui` module resolves from the
+    // omdsh installation), then the Profile's own `node_modules`, then the
+    // Profile's linked roots. No fallback links are created.
+    const resolution = await createRuntimeResolution({
         installAnchor: INSTALL_ANCHOR,
         profile: composed.profile,
-        home: omdshHome(),
+        home,
     });
-    const ctx = await boot(NAME, composed.rootConfig, structuredClone(composed.patches), (hostCtx) => {
+    const ctx = await boot(NAME, composed.rootConfig, structuredClone(composed.patches), async (hostCtx) => {
         app.current = hostCtx;
+        hostCtx.provide('profileContext', profileContext);
         hostCtx.provide(DSH_LAUNCH_ENVIRONMENT_KEY, environment);
+        await hostCtx.plugin(PluginPackages, { resolution });
         provideCmdline(hostCtx, {
             args: resume === undefined ? prompt : ['--resume', resume],
             exit: (code) => { void shutdown.shutdown(code); },

package/lib/composition.d.ts

@@ -5,7 +5,7 @@
  */
 import { type ConfigDumpLayer } from '@deepseek-ai/dsh-app-boot';
 import type { LaunchEnvironmentSnapshot } from '@deepseek-ai/dsh-launch-environment';
-export { homePatchPath, INSTALL_ANCHOR, NAME, PRODUCT_BUNDLE, PROFILE_NAME, PROFILE_PATCH_LABEL, SHIPPED_PRESET_ROOT, } from './profile.ts';
+export { homePatchPath, INSTALL_ANCHOR, NAME, PRODUCT_BUNDLE, PROFILE_NAME, PROFILE_PATCH_LABEL, } from './profile.ts';
 /**
  * Materialize the same inherited > project `.env` > home `.env` snapshot
  * that live boot uses, so dump, plugin, and mount see one process environment.

package/lib/composition.js

@@ -5,7 +5,7 @@
  */
 import { loadLayeredEnv, renderConfigDump, } from '@deepseek-ai/dsh-app-boot';
 import { composeLaunch, NAME } from "./profile.js";
-export { homePatchPath, INSTALL_ANCHOR, NAME, PRODUCT_BUNDLE, PROFILE_NAME, PROFILE_PATCH_LABEL, SHIPPED_PRESET_ROOT, } from "./profile.js";
+export { homePatchPath, INSTALL_ANCHOR, NAME, PRODUCT_BUNDLE, PROFILE_NAME, PROFILE_PATCH_LABEL, } from "./profile.js";
 /**
  * Materialize the same inherited > project `.env` > home `.env` snapshot
  * that live boot uses, so dump, plugin, and mount see one process environment.

package/lib/profile.d.ts

@@ -15,8 +15,6 @@ export declare const PROFILE_ROOT_FILENAME = "cordis.yml";
 export declare const PROFILE_PATCH_LABEL = "profiles/omdsh/cordis.patch.yml";
 /** Absolute path of this installation's package.json (src/ and lib/ both sit one level under apps/omdsh). */
 export declare const INSTALL_ANCHOR: string;
-/** Shipped agent-preset root: beside this app's own config, in both source and built layouts. */
-export declare const SHIPPED_PRESET_ROOT: string;
 /** Absolute path of the machine-local user patch layer. */
 export declare function homePatchPath(environment?: NodeJS.ProcessEnv): string;
 /** Keep the product bundle first; user bundles stay after it in their current order. */
@@ -28,8 +26,8 @@ export declare function ensureProductBundleFirst(dir: string): void;
 export declare function ensureOmdshProfile(home: string): string;
 /**
  * Ensure the Profile exists, load its bundles, and rewrite the empty root
- * include. Module fallback healing happens asynchronously in runOmdsh after
- * this function returns the resolved Profile.
+ * include. `runOmdsh` derives the runtime package resolution from the
+ * resolved Profile after this function returns.
  */
 export declare function prepareProfile(home: string, userLayer?: boolean): Profile;
 /** One launch's labeled layers and the flattened patch list `boot()` mounts. */
@@ -38,16 +36,12 @@ export interface LaunchComposition {
     rootConfig: string;
     layers: ConfigDumpLayer[];
     patches: PatchOptions[];
+    /** Layers above the Profile patch; the Profile facts re-derive them for a reconciliation. */
+    overlays: PatchOptions[];
 }
-/**
- * Restate the agent-presets row with the shipped roster root. Profile
- * `baseUrl` would otherwise resolve `./agent-presets/` inside the Profile
- * directory, which does not carry the product presets.
- */
-export declare function agentPresetsOverlay(layerPatches: readonly PatchOptions[][]): PatchOptions | undefined;
 /**
  * Compose the live launch layers in product → user bundles → Profile patch
- * → home patch → MCP → LSP → agent-presets overlay order.
+ * → home patch → MCP → LSP order.
  */
 export declare function composeLaunch(cwd?: string, environment?: NodeJS.ProcessEnv, options?: {
     userLayer?: boolean;

package/lib/profile.js

@@ -9,7 +9,7 @@
 import { existsSync, writeFileSync } from 'node:fs';
 import { join } from 'node:path';
 import { fileURLToPath } from 'node:url';
-import { composeEntries, initProfile, loadOptionalPatches, loadProfile, PROFILE_PATCH_FILENAME, readProfileManifest, resolveProfileDir, writeProfileManifest, } from '@deepseek-ai/dsh-app-boot';
+import { initProfile, loadOptionalPatches, loadProfile, PROFILE_PATCH_FILENAME, readProfileManifest, resolveProfileDir, writeProfileManifest, } from '@deepseek-ai/dsh-app-boot';
 import { omdshHome } from "./config-paths.js";
 import { loadLspPatches } from "./lsp-config.js";
 import { loadMcpPatches } from "./mcp-config.js";
@@ -20,13 +20,11 @@ export const PROFILE_ROOT_FILENAME = 'cordis.yml';
 export const PROFILE_PATCH_LABEL = 'profiles/omdsh/cordis.patch.yml';
 /** Absolute path of this installation's package.json (src/ and lib/ both sit one level under apps/omdsh). */
 export const INSTALL_ANCHOR = fileURLToPath(new URL('../package.json', import.meta.url));
-/** Shipped agent-preset root: beside this app's own config, in both source and built layouts. */
-export const SHIPPED_PRESET_ROOT = fileURLToPath(new URL('../config/agent-presets/', import.meta.url));
 /** The empty root entry list every Profile tree patches over. */
 const PROFILE_ROOT_CONFIG = `# omdsh profile root — an empty entry list. The tree is composed as patches:
 # the @agi-fans/oh-my-dsh product bundle, then each extra name in
 # package.json's dsh.profile.bundles, then cordis.patch.yml, then the home
-# patch and MCP inserts. Edit cordis.patch.yml, not this file.
+# patch, MCP inserts, and LSP inserts. Edit cordis.patch.yml, not this file.
 []
 `;
 /** Absolute path of the machine-local user patch layer. */
@@ -58,8 +56,8 @@ export function ensureOmdshProfile(home) {
 }
 /**
  * Ensure the Profile exists, load its bundles, and rewrite the empty root
- * include. Module fallback healing happens asynchronously in runOmdsh after
- * this function returns the resolved Profile.
+ * include. `runOmdsh` derives the runtime package resolution from the
+ * resolved Profile after this function returns.
  */
 export function prepareProfile(home, userLayer = true) {
     ensureOmdshProfile(home);
@@ -67,26 +65,9 @@ export function prepareProfile(home, userLayer = true) {
     writeFileSync(join(profile.dir, PROFILE_ROOT_FILENAME), PROFILE_ROOT_CONFIG);
     return profile;
 }
-/**
- * Restate the agent-presets row with the shipped roster root. Profile
- * `baseUrl` would otherwise resolve `./agent-presets/` inside the Profile
- * directory, which does not carry the product presets.
- */
-export function agentPresetsOverlay(layerPatches) {
-    const row = composeEntries([...layerPatches]).find(entry => entry.id === 'agent-presets');
-    if (row === undefined)
-        return undefined;
-    return {
-        id: 'agent-presets',
-        config: {
-            ...(row.config ?? {}),
-            roots: [{ path: SHIPPED_PRESET_ROOT, trust: 'system' }],
-        },
-    };
-}
 /**
  * Compose the live launch layers in product → user bundles → Profile patch
- * → home patch → MCP → LSP → agent-presets overlay order.
+ * → home patch → MCP → LSP order.
  */
 export function composeLaunch(cwd = process.cwd(), environment = process.env, options = {}) {
     const userLayer = options.userLayer !== false;
@@ -108,14 +89,12 @@ export function composeLaunch(cwd = process.cwd(), environment = process.env, op
     const lsp = loadLspPatches(cwd, environment);
     if (lsp.length > 0)
         layers.push({ label: 'lsp.json', patches: lsp });
-    const overlay = agentPresetsOverlay(layers.map(layer => layer.patches));
-    if (overlay !== undefined)
-        layers.push({ label: 'agent-presets', patches: [overlay] });
     return {
         profile,
         rootConfig: join(profile.dir, PROFILE_ROOT_FILENAME),
         layers,
         patches: layers.flatMap(layer => layer.patches),
+        overlays: [...mcp, ...lsp],
     };
 }
 //# sourceMappingURL=profile.js.map

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@agi-fans/oh-my-dsh",
   "description": "omdsh: a keyboard-first DeepSeek TUI coding agent on the DeepSeek Harness runtime",
-  "version": "0.17.0",
+  "version": "0.19.0",
   "homepage": "https://omdsh.agi.fans/",
   "keywords": [
     "omdsh",
@@ -41,7 +41,13 @@
   ],
   "dsh": {
     "bundle": {
-      "patch": "./config/cordis.yml"
+      "patch": [
+        "./config/cordis.yml",
+        "./config/presets/standard.patch.yml",
+        "./config/presets/code.patch.yml",
+        "./config/presets/minimal.patch.yml",
+        "./config/presets/cordis.patch.yml"
+      ]
     }
   },
   "license": "MIT",
@@ -49,126 +55,134 @@
     "access": "public"
   },
   "dependencies": {
-    "@deepseek-ai/cordis-plugin-loader": "1.0.3",
-    "@deepseek-ai/cordis-plugin-timer": "1.1.4",
-    "@deepseek-ai/dsh-agent": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-agent-default-model": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-agent-instructions": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-agent-loop": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-agent-presets": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-app-boot": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-attachment-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-authorization": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-bash-sandbox": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-cmdline": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-code-runtime-worker-thread": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-command-compact": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-command-goal": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-commands": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-compaction-basic": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-compaction-tool-result-pruner": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-cordis-host-runner": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-credentials-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-file-reference": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-file-reference-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-fs-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-fs-observation-policy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-fs-sandbox": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-goal": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-goal-round-driver": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-http-proxy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-invariants": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-jobs-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-launch-environment": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-llm": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-llm-deepseek": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-llm-pi-ai": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-llm-retry": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-lsp": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-lsp-stdio": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-mcp-client": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-output-retention": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-permission-presets": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-persona": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-plan-mode": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-pwsh-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-pwsh-sandbox": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-repeat-tool-reminder": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-sandbox-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-sandbox-policy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-scope": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-checkpoint-policy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-persistence-jsonl": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-projection": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-projection-cache": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-query": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-query-sqlite": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-reference": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-stats": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-title": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-session-title-first-prompt-llm": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-settings": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-settings-file": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-shell-env": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-skill": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-skill-filesystem": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-spill": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-spill-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-spill-policy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-storage": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-storage-domain": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-storage-json": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subagent": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subagent-acp": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subagent-fork-in-process": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subagent-in-process-driver": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subagent-spawn-in-process": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-subprocess-local": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-system-prompt": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-terminal": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-terminal-bash": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-timeout": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-token-meter": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-ask-user": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-bash": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-bash-persistent": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-call-timeout-policy": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-cordis": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-fs": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-fs-search": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-goal": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-jobs": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-lsp": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-present": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-pwsh": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-pwsh-persistent": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-ralph": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-skill": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-str-replace-editor": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-subagent": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-subagent-control": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-terminal": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-todo": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-web": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tool-workflow": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-workspace": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-tools": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-user-approval": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-user-questions": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-web": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-web-fetch-http": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-workflow": "0.1.5-rc.2",
-    "@deepseek-ai/dsh-workflow-worker-thread": "0.1.5-rc.2",
-    "@deepseek-ai/schemastery": "3.18.2",
-    "node-addon-require-builtin": "0.1.4",
-    "@agi-fans/dsh-tui": "^0.17.0"
+    "@deepseek-ai/cordis-plugin-loader": "1.0.5",
+    "@deepseek-ai/cordis-plugin-timer": "1.1.6",
+    "@deepseek-ai/dsh-agent": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-default-model": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-instructions": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-loop": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-preset": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-preset-registry": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-agent-tool-presentation": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-app-boot": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-attachment-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-authorization": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-bash-sandbox": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-cmdline": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-command-compact": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-command-feedback": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-command-goal": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-commands": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-compaction-basic": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-compaction-image-offload": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-compaction-tool-result-pruner": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-config-editor": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-cordis-host-runner": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-credentials-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-file-reference": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-file-reference-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-fs-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-fs-observation-policy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-fs-sandbox": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-goal": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-goal-round-driver": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-http-proxy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-invariants": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-jobs-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-launch-environment": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-llm": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-llm-deepseek-api-key": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-llm-pi-ai": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-llm-retry": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-lsp": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-lsp-stdio": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-mcp-client": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-mcp-resources": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-output-retention": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-permission-presets": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-persona": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-plan-mode": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-ptc-runtime-node": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-pwsh-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-pwsh-sandbox": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-repeat-tool-reminder": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-sandbox-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-sandbox-policy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-scope": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-checkpoint-policy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-persistence-jsonl": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-projection": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-projection-cache": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-query": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-query-sqlite": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-reference": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-stats": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-title": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-session-title-first-prompt-llm": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-settings": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-shell-env": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-skill": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-skill-filesystem": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-spill": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-spill-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-spill-policy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-storage": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-storage-domain": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-storage-json": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subagent": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subagent-acp": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subagent-fork-in-process": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subagent-in-process-driver": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subagent-spawn-in-process": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-subprocess-local": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-system-prompt": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-terminal": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-terminal-bash": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-timeout": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-token-meter": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-ask-user": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-bash": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-bash-persistent": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-call-timeout-policy": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-cordis": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-fs": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-fs-search": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-goal": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-jobs": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-lsp": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-present": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-pwsh": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-pwsh-persistent": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-ralph": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-skill": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-str-replace-editor": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-subagent": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-subagent-control": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-terminal": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-session-query": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-todo": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-web": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tool-workflow": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-tools": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-user-approval": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-user-questions": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-web": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-web-fetch-http": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-web-search-deepseek": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-workflow": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-workflow-ptc": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-workspace": "0.2.0-rc.2",
+    "@deepseek-ai/dsh-workspace-changes": "0.2.0-rc.2",
+    "@deepseek-ai/schemastery": "3.18.4",
+    "node-addon-require-builtin": "^0.1.6",
+    "@agi-fans/dsh-tui": "^0.19.0"
   },
   "peerDependencies": {
-    "@deepseek-ai/cordis": "^4.0.2"
+    "@deepseek-ai/cordis": "^4.0.4"
   },
   "devDependencies": {
-    "@deepseek-ai/cordis": "4.0.2",
+    "@deepseek-ai/cordis": "4.0.4",
     "@types/node": "^22.10.0",
     "tsx": "^4.19.0",
     "typescript": "^5.6.0",

package/config/agent-presets/code/agent.cordis.yml

@@ -1,5 +0,0 @@
-- id: persona
-  name: '@deepseek-ai/dsh-persona'
-  config:
-    prefix: >-
-      You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}. Prefer programmatic tool calling for multi-step operations when the generated tools SDK is available.

package/config/agent-presets/code/preset.yml

@@ -1,3 +0,0 @@
-name: PTC
-description: Standard coding agent optimized for programmatic tool calling through the generated TypeScript SDK.
-order: 2

package/config/agent-presets/cordis/agent.cordis.yml

@@ -1,10 +0,0 @@
-- id: persona
-  name: '@deepseek-ai/dsh-persona'
-  config:
-    prefix: |-
-      You are a coding agent powered by the {{model}} model, running on the DeepSeek Harness. Your working directory is {{cwd}}.
-
-      You can inspect and experiment with the live Cordis runtime. Host composition owns shared registries and backends; an Agent preset owns one session's tools, persona, prompt sections, and isolated services. Treat runtime mutation as shell-level authority and explain the affected composition boundary before changing it.
-
-- id: tool-cordis
-  name: '@deepseek-ai/dsh-tool-cordis'

package/config/agent-presets/cordis/preset.yml

@@ -1,3 +0,0 @@
-name: Cordis
-description: Standard coding agent with live runtime inspection and plugin experimentation for authoring Harness compositions.
-order: 4

package/config/agent-presets/minimal/agent.cordis.yml

@@ -1,52 +0,0 @@
-- id: persona
-  name: '@deepseek-ai/dsh-persona'
-  config:
-    prefix: You are a helpful software engineer assistant.{{omdsh_agent_behavior}}
-    complete: true
-    includeRuntimeContext: false
-
-# `shell` is an omdsh alias for the host's one shell tool (`bash` on POSIX,
-# `pwsh` on Windows), so this two-tool shape holds on every platform without a
-# name that `tools.restrict()` would reject.
-- id: profile
-  name: '@agi-fans/dsh-tui/agent-profile'
-  config:
-    tools:
-      allow:
-        - shell
-        - str_replace_editor
-
-- id: persistent-shell
-  name: cordis:group
-  group: true
-  isolate:
-    terminals: true
-  config:
-    - id: terminal
-      name: '@deepseek-ai/dsh-terminal'
-
-    - id: terminal-bash
-      name: '@deepseek-ai/dsh-terminal-bash'
-      disabled: !!js process.platform === 'win32'
-      config:
-        timeoutMs: 300000
-
-    - id: persistent-bash
-      name: '@deepseek-ai/dsh-tool-bash-persistent'
-      disabled: !!js process.platform === 'win32'
-      config:
-        timeoutMs: 300000
-
-    # The PTY backend is shared; the dialect selects PowerShell on Windows.
-    - id: terminal-pwsh
-      name: '@deepseek-ai/dsh-terminal-bash'
-      disabled: !!js process.platform !== 'win32'
-      config:
-        shellDialect: pwsh
-        timeoutMs: 300000
-
-    - id: persistent-pwsh
-      name: '@deepseek-ai/dsh-tool-pwsh-persistent'
-      disabled: !!js process.platform !== 'win32'
-      config:
-        timeoutMs: 300000

package/config/agent-presets/minimal/preset.yml

@@ -1,3 +0,0 @@
-name: Minimal
-description: Focused two-tool coding agent with a persistent Bash session and str_replace_editor.
-order: 3

package/config/agent-presets/standard/agent.cordis.yml

@@ -1,5 +0,0 @@
-- id: persona
-  name: '@deepseek-ai/dsh-persona'
-  config:
-    prefix: >-
-      You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}.

package/config/agent-presets/standard/preset.yml

@@ -1,3 +0,0 @@
-name: Standard
-description: Full coding agent with files, shell, search, skills, planning, goals, and delegation.
-order: 1