Function ci_declared_edge_count 

pub fn ci_declared_edge_count(ci: &CiRun) -> usize

Substrate-canonical per-Acao declared-edge-count projection every consumer of a borrowed canteiro_types::CiRun that needs the total number of author-declared deps edges across every canteiro_types::CiNode keys off — returns the plain usize sum ci.nodes.iter().map(|n| n.deps.len()).sum() verbatim, without running canteiro_types::decompose again (the count is a property of the borrowed run’s shape, not of the owned canteiro_types::CanteiroDag the sibling decompose_ci returns — an author-declared cycle carries the same edge count as an author-declared linear DAG of the same node-and-dep list).

The declared-edge-count axis carries the “how many deps edges did this repo’s CI author write?” projection every per-Acao consumer downstream fans on: the caixa_actions::RenderedAcao::edge_count artifact the M0 renderer’s validate returns (paired with the topological node-name list from cd.topo_order()), the deferred sui-supercacheci::canteiro::emit_gha workflow renderer’s per-workflow jobs.<job>.needs count reconciliation pass (each needs entry maps 1:1 to a deps edge, so a renderer that emits N edges must have consumed exactly declared_edge_count needs entries across the fan-out), a future feira lint --acao per-caixa admission verb’s per-repo declared-edge summary, a future M4 acao.pleme.io/v1alpha1/Acao CR materializer’s admission webhook spanning the declared edge count against a per-tenant complexity cap.

Prior to this lift the ci.nodes.iter().map(|n| n.deps.len()).sum() expression was inlined at two sites — caixa_actions::validate’s edge_count field construction at caixa-actions/src/lib.rs:159 (the M0 per-Acao renderer’s sole production consumer) and its own require_acao_view byte-parity pin at caixa-actions/src/lib.rs:735 (which reconstructs the same sum through the compound helper’s returned &CiRun to pin that the two paths agree) — two open-coded arithmetic expressions that expressed no compile-time link back to the typed canteiro_types::CiRun axis, so a future refactor of the declared-edge-count shape (a promotion of the plain usize sum to a {intra_workspace, cross_workspace} split once canteiro_types::CiNode grows a workspace-scoped edge kind, a per-:ci deps-edge-canonicalization pass that collapses duplicate edges once the canteiro-types axis grows a set-shaped deps representation, a per-env-class edge-weight overlay once the M4 EnvClass axis grows a per-edge cost model) would have had to be threaded through both open-coded copies in lockstep or the M0 renderer’s edge_count artifact would silently disagree with its own byte-parity pin. Lifting the projection to a typed method on the substrate primitive means every downstream consumer of the Acao’s declared-edge-count surface reaches for exactly one typed dispatch — the resolver’s accept-set migrates as a unit on any future axis addition.

The docstring on require_acao_view already named this expression verbatim (“the borrowed run for per-canteiro_types::CiNode axes (ci.nodes.iter().map(|n| n.deps.len()).sum() for the declared edge count …)”) but the substrate carried no primitive for it — the citation was documentation-only, and the two open-coded call sites re-expressed the arithmetic each time. This lift closes that gap: the docstring now cites the substrate primitive by name and every consumer reaches for the same ci_declared_edge_count one-liner.

Peer of the sibling require_ci / decompose_ci / require_acao_view per-Acao primitives on the substrate’s per-kind renderer entry-gate surface, extended onto the “borrowed canteiro_types::CiRun scalar projection” axis (the two prior primitives return borrowed / owned structural artifacts; this one returns a plain usize scalar over the borrowed run’s node-list shape). Same “one typed dispatch on the substrate primitive, thin projections at each consumer” discipline the peer per-Aplicacao crate::aplicacao::AplicacaoSpec::port_for_destination scalar projection carries on the per-Aplicacao :entrada port-resolution axis, extended onto the per-Acao :ci declared-edge-count axis.

Named ci_declared_edge_count (rather than declared_edge_count) to keep the substrate-side helper namespace explicit that the input axis is a :ci slot — matching the peer require_ci / decompose_ci ci_-prefix-shaped naming convention the sibling per-Acao substrate primitives already carry, so a caller reading caixa_core::ci_declared_edge_count(ci) sees the axis at the helper name rather than at a lifted-out use alias.