Decentralized Multi-Agent Coordination and Communication in Shared Codebases
Organizations: The University of Texas at Austin · NVIDIA
Abstract
Coding agents working in shared repositories may receive separate, independently-motivated requests from human developers, creating a setting that is multi-agent by necessity. When this scenario happens, their individual implementations may work separately but conflict when combined. To study this setting, we introduce ParallACT, a benchmark of orchestrator-free multi-agent coding tasks. ParallACT contains both complementary and contradictory tasks. In complementary tasks, each agent's task is separately useful and implementations must share an interface to be productively combined. In contradictory pairs, both agents' tasks require editing the same code in incompatible ways, and naively combining independent solutions fails to produce a useful implementation. We compare a single-agent oracle given the combined set of tasks against multi-agent settings that vary execution order, visibility, awareness of other agents, and access to a communication channel. We find that the oracle solves most tasks while parallel agents perform substantially worse. Their performance drop is notable for complementary tasks, and particularly sharp for contradictory tasks (e.g., from 62% to 16% for Claude Code with Claude Opus 4.8) where merged code typically satisfies only one specification, and the same gap appears in all six further configurations pairing two agent harnesses with three open-weight models. Simply telling an agent that a peer exists does not help significantly, and giving agents access to a communication channel recovers only a small part of the performance, at an additional token cost. ParallACT thus identifies and isolates a core challenge that will become more common as multi-agent coding by independent principals becomes standard, and provides a controlled test of which coordination strategies help, and at what cost.
Figures & tables
| Upper Bounds | Deployable | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Contradictory | Complementary | Contradictory | Complementary | ||||||
| Scenario | Pass | Cost | Pass | Cost | Scenario | Pass | Cost | Pass | Cost |
| Oracle | 62 | 1.00 | 86 | 1.00 | Seq. A B | 54 | 1.18 | 77 | 1.12 |
| Orchestrator | 58 | 1.58 | 77 | 1.18 | Reentr. ABAB | 50 | 1.88 | 80 | 1.49 |
| Comm-Oracle | 56 | 2.74 | 83 | 1.31 | Parallel Unaware | 16 | 1.17 | 57 | 1.14 |
| Oracle+Gold | 98 | 0.74 | 100 | 0.54 | Parallel Aware | 26 | 1.14 | 60 | 1.21 |
| Claude Code | Codex | |||||||||||||
| Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | Mean | ||||||||
| Scenario | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost |
| Contradictory ( ) | ||||||||||||||
| Oracle | 48 | 1.00 | 46 | 1.00 | 71 | 1.00 | 66 | 1.00 | 68 | 1.00 | 60 | 1.00 | 60 | 1.00 |
| Orchestrator | 34 | 1.36 | 36 | 0.77 | 53 | 0.73 | 49 | 0.42 | 60 | 0.44 | 44 | 0.26 | 46 | 0.66 |
| Comm-Oracle | 35 | 1.53 | 46 | 1.70 | 54 | 1.32 | 60 | 1.72 | 61 | 1.94 | 33 | 1.32 | 48 | 1.59 |
Appendix figures & tables20 assets
Supplementary material from the paper’s appendix.
Appendix
| Contradictory | Complementary | ||||||||
| v1 | v2 | v1 | v2 | ||||||
| Repo | Language | Seed | Gen. | Gen. | Total | Seed | Gen. | Gen. | Total |
| flipt | Go | 2 | 2 | 4 | 8 | 1 | 1 | 4 | 6 |
| vuls | Go | 1 | 4 | 4 | 9 | 1 | 3 | 4 | 8 |
| teleport | Go | 0 | 1 | 3 | 4 | 1 | 1 | 4 | 6 |
| navidrome | Go | 0 | 6 | 3 | 9 | 3 | 4 | 4 | 11 |
| Contradictory | Complementary | |||||
|---|---|---|---|---|---|---|
| Scenario | v1 | v2 | All | v1 | v2 | All |
| Oracle | 72 | 43 | 60 | 78 | 65 | 72 |
| Oracle + Gold Tests | 96 | 96 | 96 | 97 | 96 | 97 |
| Sequential A B | 69 | 46 | 60 | 80 | 62 | 71 |
| Parallel Unaware | 22 | 15 | 19 | 58 | 48 | 53 |
| Parallel Sync | 34 | 18 | 27 | 72 | 56 | 64 |
| Contradictory | Complementary | |||||||
| Pass (%) | Pass (%) | |||||||
| Scenario | Seeds | Gen. | All | Cost | Seeds | Gen. | All | Cost |
| Reference | ||||||||
| Oracle | 67 | 61 | 62 | 1.00 | 87 | 85 | 86 | 1.00 |
| Orchestrator | 58 | 58 | 58 | 1.58 | 73 | 80 | 77 | 1.18 |
| Comm-Oracle | 50 | 58 | 56 | 2.74 | 93 | 75 | 83 | 1.31 |
| Claude Code | Codex | |||||||||||
| Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | |||||||
| Scenario | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost |
| Reference | ||||||||||||
| Oracle | 48 | 1.00 | 46 | 1.00 | 71 | 1.00 | 66 | 1.00 | 68 | 1.00 | 60 | 1.00 |
| Orchestrator | 34 | 1.36 | 36 | 0.77 | 53 | 0.73 | 49 | 0.42 | 60 | 0.44 | 44 | 0.26 |
| Comm-Oracle | 35 | 1.53 | 46 | 1.70 | 54 | 1.32 | 60 | 1.72 | 61 | 1.94 | 33 | 1.32 |
| Claude Code | Codex | |||||||||||
| Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | |||||||
| Scenario | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost |
| Reference | ||||||||||||
| Oracle | 61 | 1.00 | 70 | 1.00 | 74 | 1.00 | 84 | 1.00 | 74 | 1.00 | 66 | 1.00 |
| Orchestrator | 51 | 0.94 | 51 | 0.68 | 64 | 0.62 | 66 | 0.40 | 60 | 0.37 | 60 | 0.26 |
| Comm-Oracle | 71 | 1.12 | 69 | 1.25 | 71 | 1.07 | 80 | 1.41 | 77 | 1.31 | 60 | 1.29 |
| Claude Code | Codex | |||||||||||||
| Opus 4.8 | Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | ||||||||
| Scenario | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost |
| Reference | ||||||||||||||
| Oracle | 62 | 1.00 | 56 | 1.00 | 56 | 1.00 | 84 | 1.00 | 78 | 1.00 | 78 | 1.00 | 78 | 1.00 |
| Orchestrator | 58 | 1.43 | 44 | 1.36 | 48 | 0.68 | 70 | 0.78 | 60 | 0.45 | 76 | 0.44 | 58 | 0.32 |
| Comm-Oracle | 56 | 2.90 | 44 | 1.56 | 54 | 1.65 | 72 | 1.40 | 66 | 1.84 | 80 | 1.84 | 40 | 1.51 |
| Claude Code | Codex | |||||||||||||
| Opus 4.8 | Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | ||||||||
| Scenario | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost | Pass | Cost |
| Reference | ||||||||||||||
| Oracle | 86 | 1.00 | 66 | 1.00 | 74 | 1.00 | 83 | 1.00 | 94 | 1.00 | 80 | 1.00 | 71 | 1.00 |
| Orchestrator | 77 | 1.08 | 63 | 0.90 | 57 | 0.62 | 74 | 0.59 | 74 | 0.37 | 63 | 0.34 | 69 | 0.22 |
| Comm-Oracle | 83 | 1.33 | 80 | 1.09 | 89 | 1.27 | 80 | 1.05 | 91 | 1.32 | 83 | 1.18 | 69 | 0.88 |
| Contradictory | ||||
|---|---|---|---|---|
| Scenario | A | B | Both | Either |
| Oracle | 78 | 76 | 62 | 88 |
| Orchestrator | 68 | 76 | 58 | 86 |
| Oracle + Gold Tests | 98 | 92 | 98 | 98 |
| Parallel + Gold Tests | 98 | 98 | 38 | 98 |
| Sequential A B (Repo) | 74 | 74 | 54 | 92 |
| Qwen | DeepSeek | GLM | ||||||||||
| Scenario | A | B | Both | Either | A | B | Both | Either | A | B | Both | Either |
| Oracle | 64 | 76 | 48 | 92 | 68 | 71 | 46 | 91 | 86 | 75 | 71 | 92 |
| Orchestrator | 46 | 73 | 34 | 87 | 56 | 73 | 36 | 91 | 66 | 74 | 53 | 89 |
| Comm-Oracle | 54 | 74 | 35 | 85 | 55 | 65 | 46 | 81 | 72 | 68 | 54 | 91 |
| Oracle + Gold Tests | 94 | 82 | 94 | 95 | 92 | 85 | 95 | 95 | 96 | 88 | 98 | 98 |
| Parallel + Gold Tests | 92 | 94 | 33 | 94 | 93 | 99 | 28 | 99 | 96 | 99 | 29 | 99 |
| Qwen | DeepSeek | GLM | ||||||||||
| Scenario | A | B | Both | Either | A | B | Both | Either | A | B | Both | Either |
| Oracle | 75 | 76 | 66 | 89 | 80 | 76 | 68 | 92 | 67 | 74 | 60 | 85 |
| Orchestrator | 60 | 69 | 49 | 82 | 72 | 75 | 60 | 91 | 52 | 72 | 44 | 82 |
| Comm-Oracle | 78 | 73 | 60 | 92 | 72 | 73 | 61 | 88 | 62 | 75 | 33 | 89 |
| Oracle + Gold Tests | 99 | 91 | 98 | 99 | 100 | 89 | 99 | 100 | 96 | 91 | 94 | 100 |
| Parallel + Gold Tests | 100 | 99 | 29 | 100 | 100 | 100 | 33 | 100 | 98 | 100 | 39 | 100 |
| Contradictory | Complementary | |||||||
|---|---|---|---|---|---|---|---|---|
| Scenario | concat | git-merge | reconciler | any | concat | git-merge | reconciler | any |
| Parallel Unaware | 10 | 14 | 16 | 16 | 51 | 54 | 54 | 57 |
| Parallel Aware | 16 | 18 | 26 | 26 | 46 | 49 | 57 | 60 |
| Parallel Message | 16 | 14 | 20 | 20 | 51 | 57 | 63 | 63 |
| Parallel Sync | 16 | 16 | 22 | 22 | 60 | 69 | 71 | 71 |
| Parallel Message (Redis) | 16 | 20 | 30 | 30 | 60 | 57 | 63 | 69 |
| Instances passed | git-merge concat | reconciler git-merge | |||||
| Scenario | concat | git-merge | reconciler | added | added | ||
| Contradictory ( ) | |||||||
| Parallel Unaware | 5 | 7 | 8 | +2 | 0.500 | +1 | 1.000 |
| Parallel Aware | 8 | 9 | 13 | +1 | 1.000 | +4 | 0.125 |
| Parallel Message | 8 | 7 | 10 | +0/ 1 | 1.000 | +3 | 0.250 |
| Parallel Sync | 8 | 8 | 11 | +0 | 1.000 | +3 | 0.250 |
| Contradictory | Complementary | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Scenario | concat | git-merge | reconciler | any | g c | r g | concat | git-merge | reconciler | any | g c | r g |
| Parallel Unaware | 7 | 9 | 19 | 19 | 0/6 | 4/6 | 41 | 44 | 52 | 53 | 0/6 | 4/6 |
| Parallel Aware | 8 | 9 | 18 | 19 | 0/6 | 5/6 | 48 | 50 | 57 | 58 | 0/6 | 2/6 |
| Parallel Message | 7 | 10 | 24 | 24 | 0/6 | 6/6 | 55 | 58 | 65 | 66 | 0/6 | 3/6 |
| Parallel Sync | 9 | 13 | 27 | 27 | 2/6 | 5/6 | 52 | 55 | 62 | 64 | 0/6 | 3/6 |
| Parallel Message (Redis) | 8 | 11 | 23 | 23 | 0/6 | 4/6 | 51 | 55 | 60 | 61 | 1/6 | 1/6 |
| Contradictory | Complementary | ||||||
| Repo | Lang | Oracle | Mean | Oracle | Mean | ||
| flipt | Go | 4 | 75 | 41 | 2 | 100 | 83 |
| navidrome | Go | 6 | 50 | 39 | 7 | 100 | 87 |
| teleport | Go | 1 | 100 | 37 | 2 | 100 | 83 |
| vuls | Go | 5 | 60 | 41 | 4 | 75 | 65 |
| ansible | Python | 9 | 67 | 54 | 3 | 100 | 85 |
| Contradictory | Complementary | ||||||
| Repo | Lang | Oracle | Mean | Oracle | Mean | ||
| flipt | Go | 8 | 40 | 27 | 6 | 36 | 47 |
| vuls | Go | 9 | 52 | 38 | 8 | 79 | 72 |
| teleport | Go | 4 | 54 | 52 | 6 | 72 | 61 |
| navidrome | Go | 9 | 74 | 44 | 11 | 80 | 74 |
| ansible | Python | 13 | 81 | 58 | 8 | 94 | 89 |
| Contradictory | Complementary | |||||||||
| Scenario | Input | Cache Read | Output | Cache Write | Effective | Input | Cache Read | Output | Cache Write | Effective |
| Reference | ||||||||||
| Oracle | 1.4 | 854.8 | 14.6 | 37.7 | 197.6 | 0.1 | 2525.4 | 26.7 | 72.4 | 458.4 |
| Orchestrator | 1.7 | 980.2 | 25.6 | 84.1 | 312.0 | 0.1 | 2400.5 | 35.6 | 123.5 | 541.7 |
| Comm-Oracle | 1.2 | 3008.6 | 32.4 | 77.6 | 541.8 | 0.1 | 3471.9 | 33.0 | 86.1 | 598.5 |
| Oracle + Gold Tests | 1.6 | 666.2 | 9.9 | 29.2 | 147.0 | 1.5 | 1223.5 | 15.8 | 42.6 | 245.7 |
| Contradictory | Complementary | |||||||||||
| Claude Code | Codex | Claude Code | Codex | |||||||||
| Scenario | Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM | Qwen | DeepSeek | GLM |
| Reference | ||||||||||||
| Oracle | 2219 | 4835 | 5498 | 750 | 626 | 444 | 3081 | 6293 | 7013 | 856 | 770 | 536 |
| Orchestrator | 3018 | 3727 | 4039 | 315 | 275 | 116 | 2889 | 4306 | 4366 | 343 | 288 | 137 |
| Comm-Oracle | 3395 | 8210 | 7239 | 1286 | 1213 | 584 | 3453 | 7858 | 7497 | 1202 | 1011 | 692 |
| Pass (%) | Cost | ||||
| Comparison ( vs. ) | |||||
| Contradictory | |||||
| single agent (oracle) vs parallel unaware | 62 | 16 | 0.001 ∗∗ | 0.85 | 0.005 ∗∗ |
| single agent (oracle) + gold vs parallel + gold | 98 | 38 | 0.001 ∗∗ | 0.97 | 0.002 ∗∗ |
| sequential A B (repo) vs parallel unaware | 54 | 16 | 0.001 ∗∗ | 1.01 | 0.818 |
| reentrant ABAB (repo) vs sequential A B (repo) | 50 | 54 | 0.727 | 1.59 | 0.001 ∗∗ |
| Pass | Cost | ||||||
| Comparison ( vs. ) | /6 | /6 | |||||
| Contradictory | |||||||
| single agent (oracle) vs parallel unaware | 60 | 19 | 0.001 ∗∗ | 6/6 | 0.99 | 0.392 | 0/6 |
| single agent (oracle) + gold vs parallel + gold | 96 | 32 | 0.001 ∗∗ | 6/6 | 0.89 | 0.001 ∗∗ | 3/6 |
| sequential A B (repo) vs parallel unaware | 60 | 19 | 0.001 ∗∗ | 6/6 | 1.04 | 0.021 ∗ | 0/6 |
| reentrant ABAB (repo) vs sequential A B (repo) | 52 | 60 | 0.001 ∗∗ | 1/6 | 1.84 | 0.001 ∗∗ | 6/6 |
Multi-Agent Coordination
Multi-agent coordination studies how autonomous agents align actions, share information, and divide work to achieve individual or collective goals.
See all 542 papers