Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Orpheus: Engineering Multiagent Systems via Communicating Agents
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes Orpheus, an enablement-based programming model for multiagent systems that generates a protocol adapter (BSPL semantics in Jason) so agents compute enabled messages from goals rather than reacting to each incoming message.
Problem
Engineering flexible decentralized multiagent systems: reactive programming models for protocol-based agents leave protocol state reconstruction tangled with internal logic, causing incompatibility, semantic errors, and inflexibility under protocol changes.
Approach
Generate role-specific adapters from BSPL information protocols that track local state and compute enabled messages at runtime; agent plans then query enabled messages, complete their out-parameters, and attempt to send them, decoupling sends from receptions.
Key results
- Conceptual evaluation on NetBill protocol variants shows Orpheus requires only minimal changes (e.g., deleting one parameter from a plan) where reactive Jason code needs plans added across multiple states.
- Orpheus agents work without changes under unordered communication, whereas reactive agents relying on FIFO ordering would fail.
- Logistics case study shows the adapter handles correlation of labels with wrapped items transparently in the PACKER agent.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use Orpheus enablement-based programming instead of reactive message-handler plans | Jason agents implementing flexible information protocols · NetBill protocol changes | code complexity | Helped |
| Use Orpheus adapters to compute enabled messages rather than manually maintaining protocol state | Jason agents · Flexible-NetBill | implementation changes | Helped |
| Use Orpheus with unordered message delivery instead of relying on message ordering | Orpheus agents · Unordered communication service | correctness under message reordering | Helped |
| Use Orpheus to correlate information from multiple agents and messages | PACKER agent · Logistics scenario | code complexity | Helped |
Limitations
- Evaluation is conceptual rather than empirical; no quantitative code-complexity measurements are reported.
- Enabled predicates must exactly match message schemas, so pattern errors are easy to make and hard to find.
- Relies on KQML-style tell primitive for transport, which the authors note has known limitations.
Experimental setup
Metrics: code complexity (conceptual) ease of protocol change robustness to unordered communication
Baselines: reactive programming model in Jason Kiko
Builds on: BSPL Jason Kiko AgentSpeak
Closest papers by meaning
- AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration (ICML 2026 · similarity 0.85)
- EnCompass: Enhancing Agent Programming with Search Over Program Execution Paths (NeurIPS 2025 · similarity 0.85)
- Benefits and Limitations of Communication in Multi-Agent Reasoning (ICLR 2026 · similarity 0.85)
- Assemble Your Crew: Automatic Multi-agent Communication Topology Design via Autoregressive Graph Generation (AAAI 2026 · similarity 0.85)
- LTLf Synthesis on First-Order Agent Programs in Nondeterministic Environments (AAAI 2025 · similarity 0.84)
- Stochastic Self-Organization in Multi-Agent Systems (ICLR 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.