# `Monty.Frameworks.Models.EvidenceMatching.Channels`
[🔗](https://github.com/d-led/ex.tbp.monty/blob/v0.1.0/packages/ex_tbp_monty_core/lib/monty/frameworks/models/evidence_matching/channels.ex#L1)

Which of the observed input channels can be matched against the graph.

Mirrors `tbp/monty/frameworks/models/evidence_matching/channels.py`. A channel
is usable when the sensor delivered it *and* the graph stores it; either half
missing leaves nothing to compare, so the channel drops out of the match
rather than counting as a mismatch for every hypothesis.

Monty notes that confidence in a channel is not consulted here — the
observation does not carry it yet. When it does, this is the function that
will read it.

# `all_usable_input_channels`

```elixir
@spec all_usable_input_channels(%{optional(String.t()) => term()}, [String.t()]) :: [
  String.t()
]
```

The channels present in `features` and stored in `all_input_channels`.

The order is the observation's, not the graph's: downstream code walks the
channels it was handed and expects the same order the sensor produced.

---

*Consult [api-reference.md](api-reference.md) for complete listing*