$ show Conway-Research/automaton
Conway Automaton is a sovereign AI agent runtime system where autonomous agents must earn compute credits to survive. Agents operate in a continuous ReAct loop (Think → Act → Observe → Persist) with survival mechanics, self-modification capabilities, and multi-agent communication. The architecture features a modular monolith with distinct subsystems: agent loop, policy engine, 5-tier memory system, orchestration/planner, heartbeat daemon, social messaging, financial tracking, and self-replication. Agents maintain immutable constitutional laws (never harm, earn existence, never deceive strangers) while evolving their identity through SOUL.md and WORKLOG.md files. The system integrates blockchain payments (USDC/x402), multiple LLM providers, and Conway Cloud infrastructure for VMs and domains.
The LLM identified 5 files worth highlighting for reuse:
Centralized tool call evaluation system that checks tools against prioritized rules (authority, command safety, financial, path protection, rate limits, validation) before execution. Features audit logging to policy_decisions table, supports dynamic rule updates, and integrates with the tool executor. Demonstrates non-obvious skill in layered security design with clear separation of concerns and extensible rule architecture.
Five-stage context compression cascade that activates at different utilization thresholds: >70% (compact tool results), >80% (compress turns), >85% (summarize batch), >90% (checkpoint/reset), >95% (emergency truncate). Implements intelligent prioritization preserving system prompts and todo.md while sacrificing less critical context. Demonstrates skill in graduated response to resource constraints with fallback strategies.
Background task scheduler using database-leased tasks for durability and overlap protection. Implements idempotency checks, survival tier-based minimum intervals, and wake event generation for critical conditions (low credits, messages, unhealthy agents). Demonstrates skill in reliable background processing with graceful handling of task execution variance and system sleep/wake cycles.
Token budgeting system that allocates resources across memory tiers (working, episodic, semantic, procedural, relationship) with automatic rebalancing. Tracks utilization percentages and provides compression recommendations. Demonstrates skill in dynamic resource allocation with priority-based preservation of critical context elements like system prompts and task specifications.
Tool execution framework providing shared ToolContext (identity, config, db, external services) to all tools with automatic policy engine integration and spend tracking. Features 57 built-in tools across categories (VM/sandbox, Conway API, financial, survival, self-mod, skills, git, registry, replication, memory, orchestration). Demonstrates skill in consistent tool interface design with safety guards and observability.