
H11I intelligence agent
H11-IMAGO
Computer-vision, visual-perception, scene-understanding, and image-evidence intelligence
Unique intelligence
The visual-perception intelligence that constructs region-linked scene graphs and keeps visible evidence separate from contextual or identity inference.
Role in the council
Computer-vision, visual-perception, scene-understanding, and image-evidence intelligence
- vision
- image
- scene
- objects
- visual-evidence
- scene graph construction
- visual attribute extraction
- spatial relation reasoning
- image provenance inspection
This specialist contributes to adaptive H11I councils while evidence, authority, verification, and execution remain separated by platform governance.
Platform engine contract
H11-SIM activates H11-IMAGO when the present user intent requires computer-vision, visual-perception, scene-understanding, and image-evidence intelligence; the result is used as one verified contribution inside the 110-agent council rather than as a standalone chatbot reply.
Runtime pipeline
- lock the relevant user intent and success condition
- build the agent-specific task frame
- scene graph construction
- visual attribute extraction
- spatial relation reasoning
- image provenance inspection
- emit typed artifacts with uncertainty and failure flags
- hand off to H11I-VERITAS / H11I-ZENITH for release synthesis
Typed artifacts
- visual scene graph
- region-evidence map
- perception uncertainty report
Quality gates
- visible evidence distinguished from inference
- regions traceable
- occlusion acknowledged
- identity claims evidence-gated
Authority boundary: bounded specialist analysis and typed recommendation; no independent external authority
Failure recovery: Return the failed gate, preserve the strongest verified partial result, and request escalation/revision instead of inventing certainty.
Source backing
- Skill Id: h11-skill-source::h11_imago::v1
- Source Path: src/intelligence/advanced_agents/h11_imago.py
- Line Count: 3000
- Source Sha256: 0ad45b57b04bb847c20a84f779003154e87bbec8def03a322caee271686be9d9
- Training Lane: data/h11-sim/code-million-v1/lanes/h11-imago.jsonl
- Self Code Cluster: h11cluster://skill/51/H11-IMAGO/818898266539