feat(yaml): Kubernetes/CRD-aware YAML support with semantic search

[M0Rf30]

Summary

Adds YAML language support with Kubernetes/CRD-aware semantic symbols, so aft_search, aft_outline, and aft_zoom work on manifests and Argo Workflows.

YAML was previously unsupported (only HTML/Markdown/JSON among the query-less languages). This wires LangId::Yaml end-to-end and — critically — adds yaml/yml to the semantic-index allowlist so manifests are actually embedded for aft_search.

Design: two-tier extractor

• Tier 1 (Kubernetes/CRD): any document carrying apiVersion + kind becomes one rich symbol named <namespace>/<Kind>/<name> (falls back to metadata.generateName). Generalizes to any CRD for free, since apiVersion+kind is the CRD contract.
• Tier 2 (generic YAML): plain YAML (docker-compose, CI config, Helm values.yaml) emits top-level keys as Variable symbols.
• Multi-document (---) streams handled per document; templated/un-parseable docs degrade gracefully rather than failing the file.

Semantic enrichment (the part that makes search useful)

Symbols are not raw text — the embed text is enriched with high-signal spec fields so intent queries match:

• container image / ports
• resource cpu / memory limits & requests
• volumeMounts (mountPath), replicas
• env var names (env=A,B)
• RBAC verbs / resources / apiGroups
• Argo Workflow entrypoint / templates / command / args / schedule (covers Workflow / CronWorkflow / WorkflowTemplate)

Example outputs (from the real binary):

Deployment → cpu=2 memory=1Gi mountPath=/var/data env=DATABASE_URL,LOG_LEVEL
Role       → resources=pods,services verbs=get,list,watch
Workflow   → entrypoint=main image=alpine:3.18 command=echo templates=main,print

Implementation notes

• New tree-sitter-yaml grammar dep.
• yaml_unwrap() peels tree-sitter-yaml's block_node/flow_node value wrappers (without it, sequence/mapping detection never fires and raw multi-line YAML leaks into embed text).
• yaml_flatten_value comma-joins sequences and returns empty for nested mappings (leaf scalars collected separately).
• LangId::Yaml wired across all match sites (parser, format, imports, calls, indent, inspect scanners, configure, outline, manager).

Tests

10 unit tests covering: k8s resource naming, namespace/no-namespace, multi-doc, generic fallback, empty, resource limits, env names, RBAC rules, Argo workflow, generateName fallback. All green.

Notes / open questions for maintainers

• This is a draft — happy to adjust the symbol model, SymbolKind choice (currently Class for resources), or the enrichment key set to match project conventions.
• Deep spec flattening is intentionally scoped to high-signal keys; broader/Helm-template-aware parsing is deferred.

---

^{Need help on this PR? Tag @codesmith with what you need. Autofix is disabled.}

---

Summary by cubic

Adds YAML language support with Kubernetes/CRD-aware semantic symbols so aft_search, aft_outline, and aft_zoom work on manifests and Argo Workflows. YAML files (.yaml, .yml) are now semantically indexed across the toolchain.

• New Features

  • Kubernetes/CRD-aware symbols: emit // with generateName fallback; applies to any CRD.
  • Semantic enrichment for search: image/ports, cpu/memory/storage, replicas, mountPath, env names, RBAC verbs/resources/apiGroups, and Argo Workflow entrypoint/templates/command/args/schedule.
  • Multi-document streams supported; templated/unparseable docs are skipped per document.
  • Generic YAML fallback: top-level keys become Variable symbols for non-k8s YAML.
  • End-to-end wiring: parser, outline, configure, format, indent, imports/calls/inspect; yaml/yml added to the semantic index allowlist.
  • Added 10 YAML unit tests and updated README.

• Dependencies

  • Added tree-sitter-yaml (v0.7.2).

^{Written for commit 33542af. Summary will update on new commits.}

[cubic-dev-ai]

2 issues found across 17 files

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>