Preface

The core philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When building agents or workflows, growth data is often scattered across Markdown notes, CSV exports, and JSONL logs. How to turn this fragmented data into interpretable diagnostics and actionable SOPs is a common pain point for developers. The winyh/dsh-growth plugin targets this scenario and provides a set of local-first analysis and execution tools.

What is this

winyh/dsh-growth is a local-first DeepSeek Harness plugin developed by maintainer winyh. It transforms growth materials and business data in Markdown, CSV, and JSONL formats into interpretable growth diagnostics, implementation plans, and execution SOPs.

Core features

  1. Local-first processing: The plugin relies on the local file system at runtime, and private docs/ files are excluded from Git and package contents.
  2. Multi-format parsing: Supports parsing Markdown, CSV, and JSONL formats to read growth materials and business data.
  3. Validation and attribution:
    - Provides growth_artifact_review to validate growth artifact IDs, content fingerprints, and expiration times.
    - Provides growth_attribution_review to match GEO/content measurement plans with actual metrics; it only compares observations that have baselines and does not automatically claim causality.
  4. Analytical coverage: Covers the AARRR funnel, activation, retention, referral, MRR, CAC/LTV/Payback, HADI experiments, RICE prioritization, and WBR/MBR reports.
  5. Output generation: Responsible for generating growth diagnostics, implementation plans, and execution SOPs.

Installation and enabling

  1. Environment check: Ensure the Node.js version satisfies ^22.19.0 || >=24.0.0. The plugin’s peer packages have been tested against DeepSeek Harness 0.1.5-rc.2 and Cordis 4.0.2.
  2. Run installation: Installing to default does not enable the plugin in web; you need to install it into the profile that is actually running (for example, web):
dsh plugin --profile web add github:winyh/dsh-growth
  1. Restart the profile: After installation, restart the running profile to make the plugin take effect.

Typical usage

  1. Start the DeepSeek Harness Web interface:
dsh web
  1. Run the installation command above in the Web interface.
  2. Use the plugin’s validation and attribution tools to analyze uploaded growth artifacts.
  3. Generate the corresponding growth diagnostics, implementation plans, and execution SOPs based on the analysis results.

Applicability and notes

  • Applicable scenarios: Suitable for product managers, growth hackers, or data analysts who need to transform business data in Markdown, CSV, or JSONL formats into concrete action plans.
  • Cautions:
    • The plugin runs with the permissions of the current DSH process; review the source code and license before analysis.
    • The plugin is responsible for analysis and planning; it does not treat unverified metrics as conclusions and emphasizes an evidence-based approach.
    • You must specify a profile during installation; otherwise, the plugin will not be enabled in the Web interface.

Brief ending

winyh/dsh-growth helps developers turn messy growth data into executable SOPs through a local-first analysis workflow. For more information, see GitHub.