
When you're ingesting semantic layer bundles and hit naming collisions, this handles the classification and resolution automatically. It distinguishes between harmless duplicates (same content, skip the write), expression-only changes (silent replace), and the messier cases like definitional contradictions where the same metric name means different things across sources. The disambiguation logic is opinionated: it renames all variants with domain suffixes, writes a unified wiki page explaining each, and flags for human review. The re-ingest versus same-ingest distinction is smart, treating a re-upload as user intent to replace while treating within-bundle conflicts as ambiguous. Stage 4 reconciliation adds cross-bundle sweeps and eviction decisions. The canonical pins feature lets you lock in prior decisions on contested keys.
$ npx -y skills add kaelio/ktx --skill ingest_triage --agent claude-codeInstalls into .claude/skills of the current project.
This skill is loaded in two contexts:
sl_discover, deterministic projection
output, existing project memory, or prior provenance overlaps with what the
current WorkUnit is about to write.Apply the rules below before every write that could collide with an existing artifact.
Is this the same artifact I'm producing now, or a different one with the same name? Read both. If names match and content matches (modulo whitespace): no conflict - skip the write, the prior one stands.
If content differs, is it an expression-only change (e.g. a different sql: body for the same measure name, same grain, same columns)?
Re-ingest change (expression-only): silently replace via sl_edit_source. No flag.
Sponsors