lintdiff inventory emits one complete lintdiff.inventory.v1 artifact from a
Cargo JSON-lines diagnostics stream. It runs before changed-line filtering,
repository code policy, verdict policy, or report budgets.
lintdiff inventory \
--diagnostics artifacts/clippy.jsonl \
--root . \
--analysis-command-file artifacts/analysis-command.json \
--upstream-exit-code 0 \
--out artifacts/lintdiff/inventory.json
The command file is a JSON string array, for example:
["cargo", "clippy", "--workspace", "--message-format=json"]The artifact retains every Cargo compiler-message, including observations
without spans, and records package and target producer identity, raw and
normalized diagnostic values, children, suggestions, source spans, completion
state, and exact process status when supplied. Raw zero or missing source
positions remain distinct from a normalized source position.
observation_id_v1 identifies one emitted observation, occurrence_id_v1
represents exact producer/code/message/location evidence, and
semantic_id_v1 is a line-independent candidate identity. context_id is
null until source context has been earned by a later analysis stage. Existing
lintdiff.report.v1 output and its finding fingerprint semantics are
unchanged.
The inventory schema is lintdiff.inventory.v1. Inventory is an experimental evidence contract for the diagnostic-delta work; it does not run base or head builds and does not claim that a head-only diagnostic was caused by a source change.