Getting Started with gosec
This guide walks you through compiling gosec, running diagnostics, inspecting local LLM runtime models, and executing the backlog manager.
1. Prerequisites
Before getting started, ensure you have:
- Go Toolchain: Go 1.27.1 or higher (compiled with
CGO_ENABLED=0). - Make: GNU Make for running automation recipes.
- Ollama (Optional for $0.00 Tier): Local Ollama daemon running on
http://127.0.0.1:11434. - Node.js (Optional for Web Portal): Node 20+ with npm for building the Astro portal.
2. Installation & Compilation
This compiles zero-CGo static binaries directly into
bin/:
- bin/gosec: The human developer CLI.
- bin/mcp-server: The stdio JSON-RPC 2.0 Model Context Protocol server.
Verify the installation:
3. System Health & Environment Doctor
Run the built-in system diagnostics suite:
Sample diagnostic report:
Running gosec system diagnostic checks...
=========================================
[OK] Go Compiler: go version go1.27.1 darwin/arm64
[OK] Git VCS: git version 2.48.1
[OK] Ollama Runtime: http://127.0.0.1:11434 (Installed models: qwen3-coder:30b, deepseek-r1:14b, llama3.3:70b)
All diagnostic checks passed. System is ready.
4. Pure-Go SQLite Backlog Management
gosec embeds a pure-Go SQLite ledger (local/backlog.sqlite) for task and dependency coordination:
5. Local LLM Runtime ($0.00 Tier)
Check your local Ollama connection and model catalogue:
# Check daemon connectivity
./bin/gosec llm status
# List installed local models
./bin/gosec llm list
# Check models loaded in GPU/VRAM
./bin/gosec llm ps
# Test a fast zero-cost generation
./bin/gosec llm generate "Explain Go channels in two sentences."
6. Astro Landing Portal & Cloudflare Pages
To test or deploy the unified Astro web portal and MkDocs documentation:
# Launch Astro local development server
make website-dev
# Build unified portal and docs
make website-build
# Deploy both website and docs to Cloudflare Pages
make deploy
🛡️ Next Steps: Runtime Security with ctxsec.io
Once your gosec application is running, integrate ctxsec.io to protect your AI agent tool calls with Google CEL policy hypervisors and in-kernel OS process sandboxing.