AnonyJcy/dsh-j-space

简体中文 | English

J-Space Cognition Suite SV1 native Agent Preset & standalone Cordis plugin for DeepSeek Harness (DSH).
Bringing internal thought representations, externalized workspace ledgers (.jspace/), and adaptive verification to unlock full LLM reasoning potential.

---

Current target: DSH 0.2.0; project version: 1.2.3. Registers j-space through a native bundle patch and aligns its tool configuration with that version's Web standard preset. The default composition targets a Web profile with the subagent model-selection settings plugin; see the host requirements below for custom or headless profiles.

DSH 0.2.0 adaptation highlights

• 1.2.0: native preset registration. The @deepseek-ai/dsh-agent-preset declaration in cordis.patch.yml registers the preset, with skills resolved relative to the package and no separate deployment step. Since DSH 0.1.7, $DSH_HOME/.agent-presets/ is no longer scanned; copying files with the legacy CLI does not register a preset in 0.2.0.
• 1.2.1: standard tool composition. Adds the /goal command and present deliverable tool, and sets modelSelectionSettings: true by default. The plugin-manager tool row remains disabled, matching the standard preset.
• 1.2.2: aligned defaults. Enables web_fetch alongside web_search and explicitly disables Ralph. Keeps the J-Space-specific persona and skill directory.
• 1.2.3: development checks and legacy plugin lifecycle. Adds Node.js type declarations and pnpm run typecheck; fixes the standalone plugin's file deployment, which waited for a ready event that Cordis 4 no longer emits. Native bundle registration is unchanged.

Historical adaptations for the DSH 0.1.5 persona schema and 0.1.6 dsh-workflow-ptc are recorded in CHANGELOG.md; they do not imply that the current 1.2.x release supports every older host.

🌟 Overview

dsh-j-space integrates the J-Space Cognition Suite (SV1 release, continuing V3.7 evolution) into DeepSeek Harness as a native Agent Preset.

Unlike traditional flat prompt injections, this plugin provides full agent scope isolation, multi-tier reasoning routes, externalized workspace ledgers (.jspace/), and adaptive verification across any compatible LLM model (DeepSeek, Claude, GPT, etc.).

---

📊 Empirical Capability Realization Report

Full evaluation reports by the original author:

• Current Benchmark Report: GLM-5.3-Flash-J-Space-Capability-Realization-Report
• Earlier Comparative Report (Preserved Archive): DeepSeek-V4-J-Space-Capability-Realization-Report

🔬 Benchmark 1: GLM-5.3-Flash Evaluation (Latest Report)

1. Main Benchmark Table (Accuracy Comparison)

Benchmark	GLM-5.3-Flash (Baseline)	GLM-5.3-Flash + J-Space V3.7/SV1†	GLM-5.3	Opus-5	Fable 5.1
HLE (w/ tools)	55.3	59.2	62.5	64.7	65.0
Terminal Bench 2.1	84.3	88.8	88.2	*89.1	*91.4
DeepSWE v1.1	63.4	68.0	66.9	68.8	67.4
Agents' Last Exam	26.3	30.5	28.5	31.6	—
AutomationBench (Public)	48.8	51.1	48.2	50.3	—

* Note: Terminal Bench 2.1 figures for Opus-5 and Fable 5.1 are independently measured by a third party; no official entries exist.
\† Estimated, based on limited controlled experiments.

2. Speed and Token Efficiency Table (GAIA Controlled Pair)

Metric	Factor / Improvement
Speed	1.87×
Token Efficiency	1.41×

---

🔬 Benchmark 2: DeepSeek-V4-Flash Evaluation (Historical Archive)

• Base Model: DeepSeek-V4-Flash-Vision-Exp
• Harness: DeepSeek Harness (Standard Mode)
• Methodology: Rigorous A/B Testing with and without J-Space on authoritative benchmark subsets and same-type mini-sets (Terminal-Bench 2.1: 20 medium / 10 hard; DeepSWE: 10 TypeScript / 10 Python / 10 Go / 2 JavaScript / 2 Rust; GAIA: level 1 / level 3, etc.), with identical model, environment, and sampling — only the J-Space toggle differs.
• Evaluation Dimensions: ① Accuracy / Pass Rate; ② Wall-clock & Token Efficiency.

1. Main Benchmark Table (Accuracy Comparison)

Benchmark	DeepSeek V4-Flash (Baseline)	DeepSeek V4-Flash + J-Space V3.7	GLM-5.3	Kimi-K3	Opus-4.8	Fable 5 (w/ fallback)
HLE (w/o tools)	*37.8	37.8	—	43.5	49.8	53.3
HLE (w/ tools)	*51.5	51.9	62.5	56.0	57.9	63.0
Terminal Bench 2.1	83.9	85.5	88.2	88.3	85.0	88.0
NL2Repo	57.7	60.4	58.0	58.0	69.7	—
CyberGym	75.3	77.8	84.5	80.0	78.3	83.1
DeepSWE	59.3	61.8	66.9	67.5	58.0	70.0
Toolathlon-Verified	75.9	77.4	73.0	76.5	76.2	77.9
Agents' Last Exam	27.3	28.3	28.5	27.6	25.7	23.8
AutomationBench (Public)	25.7	27.6	48.2	30.8	27.2	29.1
⭐ Average Score	56.99	58.61	64.54	60.96	58.33	62.13

* Note: HLE scores were not disclosed and follow DeepSeek V4-Flash-0731. The average covers the 7 rows where all six columns have values.

2. Speed and Token Efficiency Table

Benchmark	Wall-clock τ	Speedup	Output Tokens	Total Tokens	Score per Unit Time	Cost per Successful Task
HLE (w/o tools)	*1.02	−2%	−10%	+5%	0.98×	+5%
HLE (w/ tools)	0.88	+14%	−22%	+3%	1.15×	+2%
Terminal Bench 2.1	0.79	+27%	−28%	−3%	1.29×	−5%
DeepSWE	0.78	+28%	−28%	−3%	1.34×	−7%
Toolathlon-Verified	0.86	+16%	−25%	+2%	1.19×	+0%
AutomationBench (Public)	0.76	+32%	−31%	−5%	1.41×	−12%

** Note: For HLE (w/o tools), τ=1.02 is intentionally positive (i.e. slower) because on single-turn tasks without tools, injecting the full skill entry is net overhead. On long-horizon and multi-turn coding/agentic benchmarks (e.g. Terminal Bench, DeepSWE, AutomationBench), J-Space delivers +14% ~ +32% faster execution, cuts 28%~31% of output token redundancy, and boosts score per unit time by 1.15× ~ 1.41×.*

---

🚀 Installation (DSH 0.2.0)

Method 1: Install with the DSH plugin manager (recommended)

# Install into the Web profile you use, not an arbitrary project's node_modules
dsh plugin --profile web add @anonyjcy/dsh-j-space@latest

DSH loads cordis.patch.yml through the package's dsh.bundle.patch field; its declaration registers j-space. Running npm install / pnpm add in an ordinary project directory alone does not register the preset in a DSH profile.

Check that the installed package is 1.2.2 or newer. If that version is not yet published on npm, use the repository installation below. After replacing an installed package, restart the current DSH Web process and create a new session. Do not run the legacy install CLI as a registration step. For a custom Web profile, replace web with its actual profile name.

Method 2: Install a local repository checkout

git clone https://github.com/AnonyJcy/dsh-j-space.git
cd dsh-j-space

# Add the current package directory to the Web profile
dsh plugin --profile web add "$PWD"

This also registers the preset through the bundle patch. Skills are loaded from the package's preset/skills/j-space directory; no copy into the user preset directory is needed.

---

💡 Usage

1. In DeepSeek Harness Web UI

1. Create a new Session.
2. Select J-Space Cognition Suite in the Agent Preset dropdown.
3. Pick any compatible model (deepseek-chat, deepseek-reasoner, etc.) and start your task.

Configure model selection for spawn subagents in Web UI

Since 1.2.1, J-Space spawn subagents already set modelSelectionSettings: true; there is no need to add it manually. This enables integration with host settings, not unrestricted access to every model.

Open Plugins → Subagent → Model selection in DSH Web Settings, enable Allow agents to choose models for Subagents, and select permitted routes from the current DSH model catalog. Start a new session for the setting to take effect.

Routes come from each DSH deployment's own model catalog and authorization list; J-Space hardcodes no provider or model IDs. Select newly added routes in Subagent settings when needed. DSH fork subagents inherit the parent session's model by design.

2. Host requirements for custom / headless profiles

The default preset requires the Host to load @deepseek-ai/dsh-tool-subagent/model-selection-settings, supplied by the DSH Web application bundle. A host without this plugin fails to mount the preset rather than silently downgrading.

For such a host, explicitly compose the settings plugin or remove / disable the spawn subagent's modelSelectionSettings field in your own preset override. Follow your DSH profile's documentation for preset selection and launch arguments; do not assume the old dsh --preset j-space example still applies.

3. Legacy standalone Cordis plugin (migration reference only)

- id: j-space-plugin
  name: '@anonyjcy/dsh-j-space'
  config:
    autoDeploy: true

This composition only invokes the legacy file-deployment plugin; it is not a preset registration method for DSH 0.2.0. Use the bundle installation above for the current version.

---

🧩 Architecture & Data Flow

flowchart TD
    P[Profile Loads Bundle Patch] --> R[Register j-space Preset Declaration]
    R --> A[New Session]
    A --> B[Select j-space Preset]
    B --> D[Mount Preset Tools and Skills]
    D --> E[Agent Scope]
    E --> F1[Persona: J-Space SV1 Architecture]
    E --> F2[Tools: Full Coding & Reasoning Tools]
    E --> F3[Skill Filesystem: Mounted skills/j-space/]
    E --> F4[J-Space Suite: SKILL.md, 13 modules, 7 references, controller & adapters]
    E --> G[Session Model Route: Any Model]
    G --> H[Agent Executes J-Space Cognition Loop]
    H --> I[Task Workspace: Managed .jspace/ Ledger & Control State]

---

🛠️ Legacy file-deployment CLI (migration reference only)

These commands remain in the package and manage the legacy copy at $DSH_HOME/.agent-presets/j-space (default: ~/.dsh/.agent-presets/j-space). status / verify only check those files; they do not verify that a DSH 0.2.0 bundle is loaded or its preset registered. Check the new-session preset menu in the current Web profile instead.

node bin/cli.js install    # Deploy J-Space preset to ~/.dsh/.agent-presets/j-space
node bin/cli.js uninstall  # Cleanly remove J-Space preset
node bin/cli.js verify     # Verify integrity of installed preset files
node bin/cli.js status     # Display current installation status

---

📄 License

MIT License. See LICENSE. For the J-Space suite's source and author information, see the upstream project.

Maintenance

See CHANGELOG.md for release notes.