starsmash29/fr-os-ai-perimeter

How starsmash29/fr-os-ai-perimeter works

AI-Perimeter adds a model-assisted layer beside FR_OS: an administrator requests a firewall change or investigates an alert in the webUI; proposals are checked before review, while incident explanations use pseudonymised sensor evidence. The model client supports local and opt-in cloud providers. The README also documents optional per-device upload-byte reporting and evaluation suites, which are represented as separate capabilities. Relationships are limited to README statements and the shown source references; unsampled modules may have additional wiring.

Main entry points: pipeline.py Proposal Pipeline · service.py Assist Service · routes.py Screen Routes · runner.py Evaluation Runner

Components 25

Admin Interface

AI-Perimeter Screen src/ai_perimeter/webui/__init__.py
Screen Routes src/ai_perimeter/webui/routes.py
WebUI Server src/ai_perimeter/webui/server.py

Change Proposals

Proposal Pipeline src/ai_perimeter/orchestrator/pipeline.py
Patch Operations src/ai_perimeter/orchestrator/ops.py
Deny Rules src/ai_perimeter/orchestrator/deny.py
Intent Prompt src/ai_perimeter/orchestrator/prompt.py

Incident Insights

Sensor Evidence src/ai_perimeter/incident/evidence.py
Incident Explainer src/ai_perimeter/incident/explain.py
Upload Byte Collector src/ai_perimeter/upload/collector.py

Model Runtime

Assist Socket Client src/ai_perimeter/assist_client.py
Assist Service src/ai_perimeter/service.py
Socket Protocol src/ai_perimeter/protocol.py
Service Settings src/ai_perimeter/settings.py
Provider Client src/ai_perimeter/client.py

Evaluation

Evaluation Runner src/ai_perimeter/eval/runner.py
Intent Cases src/ai_perimeter/eval/cases.py
Generated Cases src/ai_perimeter/eval/generated.py
Network Simulator src/ai_perimeter/eval/sim.py
Incident Evaluations src/ai_perimeter/eval/incidents.py

Other components

Administrator
Model Provider
FR_OS Validator
FR_OS Apply Helper
FR_OS Sensors

Connections 33

Administrator → Screen Routes: requests or reviews README.md
AI-Perimeter Screen → Screen Routes: registers routes src/ai_perimeter/webui/__init__.py
WebUI Server → AI-Perimeter Screen: registers screen src/ai_perimeter/webui/server.py
Screen Routes → Assist Socket Client: asks service src/ai_perimeter/webui/routes.py
Assist Socket Client → Assist Service: sends socket task src/ai_perimeter/assist_client.py
Assist Socket Client → Socket Protocol: encodes request src/ai_perimeter/assist_client.py
Assist Service → Socket Protocol: handles messages src/ai_perimeter/service.py
Assist Service → Service Settings: loads settings src/ai_perimeter/service.py
Assist Service → Proposal Pipeline: builds proposal src/ai_perimeter/service.py
Assist Service → Incident Explainer: explains incident src/ai_perimeter/service.py
Assist Service → Provider Client: calls model src/ai_perimeter/service.py
Provider Client → Model Provider: sends chat request README.md
Screen Routes → Proposal Pipeline: requests proposal src/ai_perimeter/webui/routes.py
Proposal Pipeline → Intent Prompt: builds messages src/ai_perimeter/orchestrator/pipeline.py
Proposal Pipeline → Patch Operations: parses and applies src/ai_perimeter/orchestrator/pipeline.py
Proposal Pipeline → Deny Rules: checks proposal src/ai_perimeter/orchestrator/pipeline.py
Proposal Pipeline → FR_OS Validator: validates and lints src/ai_perimeter/orchestrator/pipeline.py
Administrator → FR_OS Apply Helper: approves change README.md
FR_OS Apply Helper → FR_OS Validator: validates before write
Screen Routes → Sensor Evidence: builds evidence bundle src/ai_perimeter/webui/routes.py
Sensor Evidence → FR_OS Sensors: reads sensor files src/ai_perimeter/incident/evidence.py
Incident Explainer → Sensor Evidence: uses bundle schema src/ai_perimeter/incident/explain.py
Screen Routes → Incident Explainer: requests explanation
Upload Byte Collector → FR_OS Sensors: collects byte data
Evaluation Runner → Intent Cases: loads cases src/ai_perimeter/eval/runner.py
Evaluation Runner → Generated Cases: loads generated suite src/ai_perimeter/eval/runner.py
Evaluation Runner → Incident Evaluations: runs incident suite src/ai_perimeter/eval/runner.py
Evaluation Runner → Proposal Pipeline: evaluates proposals src/ai_perimeter/eval/runner.py
Evaluation Runner → Provider Client: evaluates providers src/ai_perimeter/eval/runner.py
Intent Cases → Network Simulator: checks network behavior src/ai_perimeter/eval/cases.py
Generated Cases → Network Simulator: checks generated cases src/ai_perimeter/eval/generated.py
Incident Evaluations → Sensor Evidence: evaluates evidence src/ai_perimeter/eval/incidents.py
Incident Evaluations → Incident Explainer: evaluates explanations src/ai_perimeter/eval/incidents.py

Drawn by GitDiagram on November 12, 2026 from the repository's file tree, README and source files. Read as Markdown ·