Omni-IO Skills: harness-level composition lifts GPT-5.6's omni-input support from 40% to 100%
NationalUniversityofSingapore · hf · 2026-09-30
A National University of Singapore team released Omni-IO Skills, a plug-and-play Agent Harness that makes existing general-purpose agents omni-native via hierarchical Skills, a standardized multimodal execution interface, dependency-aware orchestration, and a persistent Asset Registry.
Problem: production capabilities of general agents are fragmented across text, images, audio, video, documents, 3D assets and code — extending a foundation model ties capability growth to costly model updates, while assembling specialists leaves procedures, dependencies, intermediate assets and cross-turn revisions uncoordinated.
Method: multi-asset workflows are represented as Declare Execution Graphs that schedule independent operations concurrently and register successful outputs for downstream and cross-turn reuse across replaceable backends. 27 Skills cover 38 representative tasks across seven artifact modalities and four capability families (understanding, generation, reasoning, retrieval).
Results: on UniM-90, the harness raises input-support rates of GPT-5.6 Sol and Claude Sonnet 5 from 40.00%/38.89% to 100%, with relative Semantic-Quality Coupled Score rising from 26.99/27.82 to 74.94/77.78, and Strict Structure Score hitting 100.00/99.78.
The takeaway: harness-level capability composition is a practical route to broad, evolvable Omni systems without touching the host agent's reasoning core.
More from coding & agent
- One-shots are cool, but months of AI-assisted iteration is what excites this dev — TAbrodi · 2026-09-30
- Luna finished a neural rendering client job using just 6-7% of quota — MickeySteamboat · 2026-09-30
- Adding a Skill from the Skill Hub made the same chart thesis-ready — iamfakhrealam · 2026-09-30
- Closing the laptop mid-task: a hands-on run of KooKo Agent on thesis work — iamfakhrealam · 2026-09-30
- Training an AI agent on its own explanations improves coding—no teacher, no verifier, no RL — CatAstro_Piyush · 2026-09-30
- ehartford submits With, a programming language designed for both humans and coding agents — QuixiAI · 2026-09-30