version: "1.0.0" name: frontier-calibration description: Use when Frontier simulator results must be compared with vLLM-BS ground truth, to collect a reproducible case, gate E2E metrics against the 10% threshold, find the source of an error, or apply and republish a decided Frontier correction.

Frontier Calibration

Calibration runs Frontier and vLLM-BS, an instrumented fork of vLLM v0.10.2, on the same request trace, gates the clean E2E metrics, and traces each gap to a source. The repository harness tests/comparison/e2e_gate/ decides the gate; this skill routes the work around it and constrains what may change.

If site.local.md exists beside this file, read it first. It binds the placeholders used here (<vLLM-BS checkout>, the artifacts root given as --artifacts-root, $FRONTIER_TMP_ROOT, the GPU platform) to this host and is not tracked. Without it, ask the user for those values.

Read Before a Case

1. references/contract.md: case layout and binding, ground-truth checkout,

run modes, the E2E gate, error-source classes, decisions.

1. references/call-graph.md: execution order, predicates, stops, and return

edges. It overrides any order implied elsewhere.

1. references/pitfalls.md: failures that earlier calibration tasks hit.
2. references/evidence-tiers.md: which artifact may support which claim.

Read on demand:

• references/tool-boundaries.md before launching or writing a tool;
• references/repair-approval.md before any code change;
• references/registry.json for an entry's inputs, outputs, and stop statuses

(references/registry.md explains its fields);

• references/migration.md only for records written under older entry names.

Resolving an Entry Name

Every entry is a document under entries/, not a separate skill and not a slash command. A name such as parity-run maps to entries/parity-run.md with no exception. Entry documents refer to shared material as ../references/<file>.

Route by Goal

The call graph reaches the other ten entries: groundtruth-run-check, groundtruth-run, simulator-run, dispatch-align-trace, semantic-alignment-repair, workflow-repair, op-repair, residual-repair, cpu-overhead-supplement, and cpu-overhead-repair.

Rules for Every Entry

• Only clean-run rows of the formal requests decide the gate. Each metric

passes on its own when its relative error is at most 0.10, and run_gate.py gate gives the verdict. Instrumented and diagnostic runs never close it.

• A field the user has not decided stops the work. Gather evidence, prepare

options with costs, ask one focused question, and record the user's verbatim answer under a decision ID in the task's requirements.md. No artifact, status, or earlier receipt approves a code change by itself.

• Fix simulation_logic errors first. Re-collecting or tuning measurement

tables runs only when the user approves it for the task, and a Frontier change never models a ground-truth defect.

• One correction is one ticket and one scoped commit, followed by the

regression set and fresh evidence (references/repair-approval.md).

• Published tables, readings, and run directories are never edited; a new

generation replaces them and the old one stays on record.

• GPU jobs follow the operator's platform rules. Pushing Frontier or vLLM-BS

branches and any other external action needs the user's approval.

• Write "code change" and "repair change" in records, in plain words.