Model Context Protocol (MCP) Tools Catalog

gosec exposes its entire tool ecosystem to Google Antigravity and AI coding agents via a multiplexed stdio JSON-RPC 2.0 server (bin/mcp-server).

🛠️ Backlog Management Tools (`backlog_*`)

Pure-Go SQLite task DAG coordination stored in db/backlog/backlog.sqlite:

Tool Name	Description	Key Arguments
`backlog_list_tasks`	Lists all tasks in the SQLite database	`stage`, `priority`
`backlog_get_ready_tasks`	Retrieves unblocked tasks whose DAG prerequisites are satisfied	—
`backlog_get_task`	Returns complete task record, subtasks, and dependencies	`id`
`backlog_create_task`	Registers a new task into the SQLite ledger	`title`, `description`, `priority`, `depends_on`
`backlog_update_stage`	Advances a task along the 6-phase engineering lifecycle	`id`, `stage`
`backlog_update_task`	Updates task title, description, or priority	`id`, `title`, `description`, `priority`
`backlog_add_subtask`	Appends a subtask to an existing task	`task_id`, `title`
`backlog_update_subtask`	Marks a subtask completed or updates title	`id`, `completed`
`backlog_stats`	Returns aggregate metrics across backlog stages	—

🧠 Local LLM Inference Tools (`ollama_*`)

Zero-cost ($0.00) local Ollama inference tier:

Tool Name	Description	Key Arguments
`ollama_status`	Checks local Ollama daemon connectivity	—
`ollama_list_models`	Lists all models installed locally in Ollama	—
`ollama_running_models`	Lists models currently loaded into VRAM/RAM	—
`ollama_generate`	Generates a completion using a local model ($0.00 tier)	`model`, `prompt`, `system`

🩺 System Diagnostics & Failure Triage Tools

Tool Name	Description	Key Arguments
`doctor_diagnose`	Executes comprehensive system health and compiler checks	—
`triage_log`	Analyzes CI/CD failure logs and generates structured remediation reports	`log_content`, `workflow_name`, `run_id`

🔌 Antigravity & IDE Integration

Configure Antigravity IDE or CLI (agy) by referencing bin/mcp-server in your mcp_config.json:

{
  "mcpServers": {
    "gosec": {
      "command": "./bin/mcp-server",
      "args": []
    }
  }
}

🛡️ Securing MCP Tools with ctxsec.io

When running AI agents in enterprise environments, wrap your MCP server with ctxsec.io:

flowchart LR
    Agent["Antigravity / AI Agent"] -->|JSON-RPC Request| Proxy["ctxsec.io Transparent Proxy"]
    Proxy -->|Google CEL Policy Check| Hypervisor{"Allowed by Policy?"}
    Hypervisor -->|Yes| MCPServer["bin/mcp-server (gosec)"]
    Hypervisor -->|No| Block["Block & Alert Agent"]
    MCPServer --> Proxy
    Proxy --> Agent

Hold "Alt" / "Option" to enable pan & zoom

Declarative CEL Hypervisor: Ensures only authorized tasks can be modified or executed.
Context DLP: Sanitizes sensitive tokens, API keys, and environment variables.
Deterministic Audit Log: Records all MCP operations into tamper-evident logs.