fix(validate): only enforce promoted hard checks on output-reachable nodes (BE-3406)

[mattmillerai]

ELI-5

ComfyUI's server only checks the nodes that actually feed into an output (like Save Image), plus everything upstream of them. Any node left disconnected — dangling in an exported workflow — is quietly thrown away and never checked. But PR #551 added strict "you're missing a required input" / "value out of range" rejections that fired on every recognized node, including those disconnected ones. So validate would hard-reject a workflow the server would happily run (a disconnected node especially loses its required link inputs, which the frontend only serializes when connected). This PR teaches validate to do the same pruning the server does: figure out which nodes are reachable from an output, and only apply those promoted hard checks to them.

What changed

• New helper _output_reachable_node_ids(workflow, graph): seeds from every output node (is_output_node) and walks input link references ([node_id, output_index]) backward — the same link predicate the rest of validate_workflow uses — to compute the transitive-ancestor set the server would validate. Cycle-safe (dedupes via the reachable set); dangling references (node absent from the workflow) simply aren't traversed.
• validate_workflow restricts the fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551-promoted hard checks to reachable nodes:
  • required_input_missing (_check_required_present) and autogrow_no_slots (_check_autogrow_required) are skipped for unreachable nodes.
  • below_min / above_max are demoted from hard errors back to advisory warnings on unreachable nodes (matching pre-fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551 behavior for those nodes), via a new keyword-only range_is_error arg on _validate_catalog_value.
• Left as-is (out of scope, per ticket): edge/shape/enum checks (shape_mismatch, unknown_enum_value, dangling_edge, output_index_out_of_range, edge_type_mismatch, autogrow_bare_input). These predate fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551 and were not flagged as over-rejecting; scoping only the newly-promoted checks keeps the diff minimal and avoids broadening behavior the reviewers didn't raise.

Tests

New (in TestValidateServerParity):

• test_disconnected_node_missing_required_is_pruned — acceptance (a): disconnected node missing every required input, alongside a valid connected chain → valid: true.
• test_reachable_node_missing_required_still_errors — acceptance (b): node on the output chain missing a required input still hard-errors.
• test_transitive_ancestor_missing_required_still_errors — a two-hop-upstream ancestor is reachable and still validated (proves the backward walk follows edges, not just direct parents).
• test_output_node_missing_required_still_errors — an output node itself seeds the reachable set.
• test_disconnected_out_of_range_demoted_to_warning — out-of-range on a pruned node is a warning, not a hard error; chain stays valid.
• test_reachable_out_of_range_still_errors — the connected node still hard-errors on out-of-range.

Updated: test_below_min_error, test_above_max_error, test_required_autogrow_with_no_slots_errors (from #551) each had a single non-output node with no output chain — the server would prune those too (they'd hit prompt_no_outputs, never range/required-checked). Wired the tested node to a SaveImage so the promotion is asserted where the server actually enforces it.

uv run --extra dev pytest tests/comfy_cli/cql/ → 152 passed. ruff check + ruff format --check clean.

Notes / judgment calls

• Stacked on fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551 (matt/be-3357-validate-required-inputs, base of this PR) — the promoted hard checks this PR scopes only exist on that branch, not on main. GitHub will retarget this PR to main when fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551 merges.
• Scoping decision (above): only the fix(validate): presence-check required inputs, no-outputs check, hard-error range violations (BE-3357) #551-promoted hard checks are gated on reachability; pre-existing hard checks are untouched. autogrow_bare_input (a mis-wiring/shape check, distinct from the required-_check_autogrow_required no-slots check) is likewise left as-is.

Deferred from a review thread on #551 (BE-3406).

[coderabbitai]

Warning

Review limit reached

You’ve reached a temporary PR review limit under our Fair Usage Limits Policy.

Your recent review volume is higher than typical usage, so adaptive limits are currently applied.

Next review available in: 16 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 69f7c702-45a8-459b-9217-d9c0aa517dff

📥 Commits

Reviewing files that changed from the base of the PR and between 707963d and 4295f0c.

📒 Files selected for processing (2)

• comfy_cli/cql/engine.py
• tests/comfy_cli/cql/test_engine.py

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch matt/be-3406-validate-output-reachable

✨ Simplify code

• Create PR with simplified code
• Commit simplified code in branch matt/be-3406-validate-output-reachable

---

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

🔍 Cursor Review — Consolidated panel

Triggered by @mattmillerai.

Found 4 finding(s).

Severity	Count
🟠 High	1
🟡 Medium	1
🟢 Low	2

Panel: 6/8 reviewers contributed findings.

Reviewers that did not contribute: kimi-k2.5:adversarial (empty), kimi-k2.5:edge-case (empty)

[mattmillerai]

🤖 The reviews loop filed Linear follow-up ticket(s) for review thread(s) deferred as out of scope for this PR:

• https://linear.app/comfyorg/issue/BE-3440|BE-3440 — Extend output-reachability gating to enum/class_type/edge/shape checks in validate_workflow (BE-3406 follow-up)