AgentOpsSec is a stack of small standalone tools. The stack CLI is the one entry point for setup, scanning, monitoring, and wrapped agent runs — without coupling the tools to each other.
It does not import code from the other projects. It detects which tools are installed, runs them as commands, and connects them through local JSON reports and log files. Every project still works on its own.
One install, one config (.agentopssec/config.json), and a strongest-available wrapper chain when you start an agent.
One command. No account. No telemetry.
Pick a profile during init — it sets which tools are enabled out of the box.
--profile minimalsafe--profile securityrisk--profile observeinfo--profile fullsafe--profile customwarnStack picks the strongest available wrapper chain for the agent you launch.
Stack does not import other tools — it talks to them through these local files. Run `agentopssec contract` for the active layout.
Route normal agent commands through agentopssec start, transparently.
| Set up a project. Profiles: minimal, security, observe, full, custom. |
| What is installed and what is enabled. |
| Run the available scan steps (Doctor → Firewall → Radar). |
| Local audit snapshot. |
| CI-gate variant — exits non-zero on risk or budget breach. |
| Passive snapshot of stack signals in this repo. |
| Wrap an agent with the strongest available chain. |
| Use Agent Sandbox in the chain instead of running in-place. |
| Print a shell snippet that routes codex / claude / gemini through start. |
| Run Agent Review over the relevant artifacts. |
| Pass-through to Agent Cost Lens. |
| Print the active project data contract. |
| Re-check tools, recreate folders, validate JSON/JSONL artifacts. |
| Update the stack and every installed sub-tool. |