DYSC Agent (Hybrid CLI + Runtime)

DYSC is a non-web coding agent scaffold with:

• Multi-provider API setup with primary provider selection
• Primary workspace folder management
• Skills registry (built-in + additional local installs)
• Claude-Code-style security review and AI fix suggestion flows
• Real-time dependency/context snapshot (manifests + Python packages)
• Green-signal startup readiness checks
• Local backend chat persistence with SQLite

Architecture

• Node CLI: apps/cli/main.js
• Python runtime: apps/agent-runtime/main.py
• Runtime modules: apps/agent-runtime/dysc_runtime/
• Config files: config/
• Persistent data: data/
• Skills files: skills/

Quick Start

Install like a global CLI

Install globally from the project root:

npm install -g .

Or run the installer script:

powershell -ExecutionPolicy Bypass -File scripts/install.ps1

Then run DYSC directly:

dysc health
dysc start

One-command onboarding:

dysc onboard

1) Initialize project state

node apps/cli/main.js init

This creates default config and local DB schema.

2) Configure providers (single or multiple)

List providers:

node apps/cli/main.js provider list

Add a provider:

node apps/cli/main.js provider add --id groq-main --type provider_specific --base-url https://api.groq.com/openai/v1 --api-key-env GROQ_API_KEY --enabled true

Set primary provider:

dysc provider set-primary groq-main

Secure key mapping (no plaintext key stored by DYSC):

dysc provider set-key-env groq-main GROQ_API_KEY
dysc provider key-status groq-main

Set the actual key in your OS/user env:

setx GROQ_API_KEY "<your_key_here>"

3) Set primary workspace folder

dysc workspace set D:/Spicyepanda-24-26/xystocode26/AI-26/DYSC-Agent-

Open a local project and set it as active workspace:

dysc workspace open D:/path/to/your/local-project

Use current terminal folder as active workspace:

dysc workspace use-current

Default behavior: if configured workspace is missing/invalid, DYSC auto-falls back to current project path.

4) Manage skills

List skills:

dysc skills list

Enable a skill:

dysc skills enable builtin.security-review

Install an additional local skill JSON:

dysc skills install-local custom.fixstyle D:/Spicyepanda-24-26/xystocode26/AI-26/DYSC-Agent-/skills-imports/custom.fixstyle.json

Security note: local skill imports are restricted to the skills-imports/ folder.

5) Start DYSC agent with green signal

dysc start

If all checks pass, DYSC prints:

🟢 DYSC READY

The process remains active until you stop it (Ctrl + C).

For one-time readiness verification:

dysc start --once

6) Real-time context and security review

Get current dependency/runtime context:

dysc context packages

Run automated security review:

dysc review security --limit 200

Generate a human-robust fix plan for a finding:

dysc fix suggest --file apps/agent-runtime/main.py --line 120 --rule PY-EVAL-001 --snippet "eval(user_input)"

Settings commands:

dysc settings show
dysc settings set default_model llama3
dysc settings set max_tool_rounds 6

7) Save and read chat history (backend SQLite)

Save message:

dysc chat save --session demo --role user --content "scan this repo"

List session messages:

dysc chat list demo

In-session slash commands (inside dysc start):

/help /health /review [limit] /context /settings /providers /workspace /skills /exit

Notes

• DYSC supports both OpenAI-compatible and provider-specific adapters in config.
• API keys are referenced by environment variable names (no plaintext secrets in config).
• Offline mode works with local endpoints (example: Ollama-compatible base URL).