How yogeshwars-cys/autonomous-neighbourhood-watch-anw- works
ANW is a research system for decentralized neighbourhood awareness, not a conventional centralized IDS. An operator starts an agent; each agent observes a local sensor, updates its local reasoning, exchanges information with peers, adjusts confidence, and decides whether action is warranted. The richer anw/src tree adds semantic event detection, bounded memory, world-model inference, planning, and reflection. The README documents the broad agent cycle; the sampled source confirms selected runtime entry points and capabilities, but does not expose enough calls to assert most internal subsystem wiring.
Main entry points: agent.go Agent loop · reflection_agent.go Reflection agent · communicator.go UDP communicator · main.go Agent command
Components 25
Agent runtime
- Agent command
anw/src/cmd/agent/main.go - Agent loop
anw/src/internal/agent/agent.go - Local sensor
anw/src/internal/agent/sensor.go - Agent state
anw/src/internal/agent/state.go
Local reasoning
- Event detection
anw/src/internal/agent/event.go - Anomaly detector
anw/src/internal/agent/detector.go - Event features
anw/src/internal/agent/feature.go - Episodic memory
anw/src/internal/agent/episodic_memory.go - Long-term memory
anw/src/internal/agent/long_term_memory.go - Local world model
anw/src/internal/agent/world.go - Bounded inference
anw/src/internal/agent/inference.go
Peer collaboration
- UDP communicator
anw/src/internal/agent/communicator.go - Neighbor tracking
anw/src/internal/agent/neighbor.go - Peer discovery
anw/src/internal/agent/addressbook.go - Cooperative scoring
anw/src/internal/agent/cooperation.go - Peer reputation
anw/src/internal/agent/reputation.go - Corroboration
anw/src/internal/agent/corroboration.go
Adaptive autonomy
- Advisories
anw/src/internal/agent/advisory.go - Commitment planner
anw/src/internal/agent/planner.go - Plans and actions
anw/src/internal/agent/plan_action.go - Reflection
anw/src/internal/agent/reflection.go - Reflection agent
anw/src/internal/agent/reflection_agent.go - Reflection metrics
anw/src/internal/agent/reflection_metrics.go
Other components
- Operator
- Peer agents
Connections 25
- Operator → Agent command: starts
anw/src/cmd/agent/main.go - Agent command → Agent loop: runs
anw/src/cmd/agent/main.go - Agent loop → Local sensor: observes
- Agent loop → Agent state: updates
- Agent loop → UDP communicator: exchanges updates
- UDP communicator → Peer agents: sends heartbeats
anw/src/internal/agent/communicator.go - Peer agents → UDP communicator: sends heartbeats
anw/src/internal/agent/communicator.go - Agent loop → Anomaly detector: assesses
- Anomaly detector → Event detection: classifies
- Event detection → Event features: uses features
- Agent loop → Episodic memory: records events
- Agent loop → Long-term memory: updates summaries
- Agent loop → Neighbor tracking: tracks peers
- Agent loop → Peer discovery: discovers peers
- Agent loop → Cooperative scoring: cooperates
- Cooperative scoring → Peer reputation: reinforces trust
anw/src/internal/agent/cooperation.go - Agent loop → Local world model: maintains local model
- Local world model → Bounded inference: supports inference
- Agent loop → Corroboration: corroborates
- Agent loop → Advisories: forms advisories
- Agent loop → Commitment planner: plans commitments
- Commitment planner → Plans and actions: selects actions
- Agent loop → Reflection agent: reflects
- Reflection agent → Reflection: evaluates commitments
- Reflection agent → Reflection metrics: reports measures