Qualify RSI trials with explicit completeness and lifecycle checks

docs/campaign-summary-chart.md

@@ -38,13 +38,22 @@ for the story of the accepted steps, the branch table for the full search.
 `inferrsi report` (and `run`, which calls it) renders
 `<run-dir>/summary-chart.svg` and inlines it into `report.html`. Nothing extra
 is needed for a single pairing: the figure shows the recorded baseline and
-candidate medians as a two-point chain.
+candidate medians as a two-point chain. That derived figure is withheld when
+required trial checks are unqualified (`evaluation.json` has
+`trials_qualified: false`). The report states that the retained medians are
+raw measurements, not an accepted gain. The medians themselves stay in the
+evaluation record.
 
 A campaign with several steps commits its own chain specification as
-`summary-chart.json` in the run directory. That specification wins over the
-two-point default, because it records the whole campaign rather than one
-pairing. When neither exists, the report states why no figure was drawn
-instead of showing an empty chart.
+`summary-chart.json` in the run directory. That file is historical evidence of
+the campaign, not a derivation of the current pairing, and it wins over the
+two-point default when this run may show a gain figure. The same unqualified
+check is applied before that path. The report does not write or inline an SVG,
+and it states that the committed specification is separate historical evidence
+and was not drawn for this step. The JSON file is left unchanged. Render it
+with the command below when the campaign record itself is the subject. When
+neither a qualified pairing nor a campaign specification is available, the
+report states why no figure was drawn instead of showing an empty chart.
 
 ## Use the command line for a campaign figure
 

docs/contributing.md

@@ -263,3 +263,7 @@ template's copy, validation and catalog-registration steps. Missing hardware
 verification remains explicit. Workbench experiments use the
 [quickstart](quickstart.md) and [evidence template](../evidence/_template/);
 the four CI lanes retain their current roles.
+
+For feedback-coupled models, stochastic audio and streaming/graph experiments,
+use the [Omni experiment contract](../src/inferrsi/data/skill_library/workflows/rsi-evidence/references/omni-experiments.md) to select
+reference, preflight and executable per-trial qualification checks.

docs/operations.md

@@ -223,5 +223,5 @@ unvalidated candidates.
 - **Scope:** recorded values only. The renderer computes gains, validates the specification and writes a figure; it measures nothing.
 - **Output:** an SVG figure, plus a Markdown table with the same values when `--markdown` is given. `report` and `run` also write `summary-chart.svg` into the run directory and inline it in the HTML report.
 - **Side effects:** writes the requested files; nothing is executed and no network or GPU is touched.
-- **Failure behavior:** exit code 2 with the offending field name when the specification is incomplete, ambiguous or inconsistent; a report without a paired measurement states why it has no figure.
+- **Failure behavior:** exit code 2 with the offending field name when the specification is incomplete, ambiguous or inconsistent; a report without a paired measurement, or with unqualified required trial checks, states why it has no gain figure. An unqualified run leaves a committed `summary-chart.json` undrawn, so that campaign record stays separate historical evidence and is not presented as this step's accepted gain.
 - **Example:** [campaign summary chart](campaign-summary-chart.md)

docs/quickstart.md

@@ -93,6 +93,7 @@ plan/run: all doctor flags, plus
   --candidate-repo --scenario --mode --agent --executor local
   --baseline-command --candidate-command --validation-command
   --metric-key --metric-unit --comparison-scope --direction
+  --require-trial-check (repeatable dotted boolean qualification key)
   --repetitions --min-improvement-pct --max-minutes
   --workload-id --model --model-revision --output-dir --dry-run
 status/report: --run-id --output-dir
@@ -176,3 +177,7 @@ or `--tier peer`. Export one source explicitly, for example
 `--repo-type` names the source owner; vllm_omni queries do not automatically
 include its dependencies or peers. See the
 [three-tier guide](../src/inferrsi/data/knowledge_packs/README.md) for all sources.
+
+For per-trial completeness, lifecycle and graph qualification, see the
+[Omni experiment contract](../src/inferrsi/data/skill_library/workflows/rsi-evidence/references/omni-experiments.md). These optional checks
+are frozen in RunSpec and do not replace the quality command.

docs/skill-coverage.md

@@ -4,7 +4,7 @@ Generated from the registry, the skill sidecars and the two maintained
 matrices. **Do not edit by hand**; regenerate after any registry, skill,
 matrix or frozen-plan change.
 
-- Input digest: `sha256:7640b171256091efd1514a33d5e434eef3feaed238fc92fe3eeba02fee94e9ff`
+- Input digest: `sha256:c995b02f0267f1904f6f53742b1ace9f170fc6063d030b0bbcdd8e700c0d924c`
 - Regenerate: `python tools/build_skill_coverage.py`
 - Verify without writing: `python tools/build_skill_coverage.py --check`
 
@@ -153,7 +153,7 @@ optimization is fully covered or that any procedure has measured benefit.
 | [competitor-scan](../src/inferrsi/data/skill_library/workflows/competitor-scan/SKILL.md) | common | Structure a lightweight competitive scan without inventing benchmark numbers or campaign verdicts. | Qualitative positioning of a new RSI repository against other projects before acceptance ... | failure_modes, input_contract, produced_artifact, refusal_point<br>3 advice | N/A: SKILL.md §§ Frame, Rules and Deliverable produce a qualitative competitive comparison with evidence limits, not a hardware-resource decision or implementation mechanism. | — | candidate · not_run · 0 readset(s) | no-Reference-Matrix-row, no-eval |
 | [draft-acceptance-issue](../src/inferrsi/data/skill_library/workflows/draft-acceptance-issue/SKILL.md) | common | Draft a campaign acceptance issue with checklist, RunSpec, evidence paths, and placeholders for baseline/candidate SHAs. | Opening a new acceptance campaign in an RSI-style repository, before any baseline or ... | failure_modes, input_contract, produced_artifact, refusal_point<br>3 advice | N/A: SKILL.md §§ Steps and Output prepare an acceptance issue, RunSpec, frozen gates and evidence paths; device identifiers are campaign metadata rather than a hardware-resource mechanism. | — | candidate · not_run · 0 readset(s) | no-Reference-Matrix-row, no-eval |
 | [npu-skill-factory](../src/inferrsi/data/skill_library/workflows/npu-skill-factory/SKILL.md) | common | Design and operation of the InferRSI NPU skill factory: artifact lifecycle, skill evaluation contract, corpus ... | InferRSI skill lifecycle management | failure_modes, input_contract, produced_artifact, refusal_point<br>3 advice | N/A: SKILL.md §§ Choose the evidence product, Artifact lifecycle, and What this skill does not do define skill-distillation/evaluation methodology; NPU task tiers are evaluation prerequisites, not a hardware-resource mechanism. | 2 · Methodology and corpus references (Dong1017/OmniRSI @ `codex/skill-eval-bootstrap` (82218b4))<br>2 · Methodology and corpus references (InferMatrixLab/InferRSI PR #9 @ 4ad551a9)<br>2 · Methodology and corpus references (VectorSpaceLab/AREX-Skill @ ac3fe1af)<br>2 · Methodology and corpus references (microsoft/SkillOpt @ 343db22) | candidate · not_run · 14 readset(s) | no-eval |
-| [rsi-evidence](../src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md) | common | Guide agents to publish and cite RSI evidence using plan/results/patch/validator paths and PASS/FAIL/INCONCLUSIVE ... | Documenting or reviewing an optimization campaign whose verdict will be cited downstream | failure_modes, input_contract, produced_artifact, refusal_point<br>4 advice | 1 Accelerator · Verify | — | candidate · not_run · 1 readset(s) | no-Reference-Matrix-row, no-eval |
+| [rsi-evidence](../src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md) | common | Guide agents to publish and cite RSI evidence using plan/results/patch/validator paths and PASS/FAIL/INCONCLUSIVE ... | Documenting or reviewing an optimization campaign whose verdict will be cited downstream | failure_modes, input_contract, produced_artifact, refusal_point<br>5 advice | 1 Accelerator · Verify | — | candidate · not_run · 1 readset(s) | no-Reference-Matrix-row, no-eval |
 
 ## Documents outside the registry
 

src/inferrsi/arguments.py

@@ -86,6 +86,10 @@ def parser() -> argparse.ArgumentParser:
             default="metrics.latency_ms",
             help="Flat or dotted key in benchmark JSON",
         )
+        child.add_argument(
+            "--require-trial-check", action="append", default=[], metavar="DOTTED_KEY",
+            help="Require a literal true qualification field in every benchmark JSON; repeatable",
+        )
         child.add_argument("--metric-unit", default="ms")
         child.add_argument("--comparison-scope", default="external_command")
         child.add_argument("--direction", choices=("minimize", "maximize"), default="minimize")
@@ -291,6 +295,7 @@ def command(value: str | None) -> tuple[str, ...] | None:
         candidate_command=command(args.candidate_command),
         validation_command=command(args.validation_command),
         metric_key=args.metric_key,
+        required_trial_checks=tuple(args.require_trial_check),
         metric_unit=args.metric_unit,
         comparison_scope=args.comparison_scope,
         direction=args.direction,

src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md

@@ -15,7 +15,7 @@ These are knowledge abstracts (`--kind knowledge`), with namespaced package
 IDs distinct from explicitly selected `checkout_ref` skills. Ascend/CUDA
 procedures use their corresponding backend; shared methods use `common`.
 
-Source snapshot: [7aa3344bf84c42467f43b3a927e1c7d13642bd55](https://github.com/InferMatrixLab/InferRSI/tree/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library).
+Source snapshot: [9e6a51cab505a8ae17575dc83ea53fc0d3b8b027](https://github.com/InferMatrixLab/InferRSI/tree/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library).
 
 ```bash
 inferrsi knowledge search 'profiling' --repo-type inferrsi --limit 50
@@ -29,7 +29,7 @@ contain the current input paths and bytes (Git checkout filters are respected).
 
 ```bash
 python tools/build_skill_index.py --source-revision <committed-source-revision>
-python tools/build_skill_index.py --source-revision 7aa3344bf84c42467f43b3a927e1c7d13642bd55 --check
+python tools/build_skill_index.py --source-revision 9e6a51cab505a8ae17575dc83ea53fc0d3b8b027 --check
 ```
 
 `--check` detects missing, stale and unexpected generated records without writing.

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json

@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json

@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json

@@ -27,8 +27,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json

@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json

@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json

@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json

@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json

@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json

@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-video-diffusion-optimization.json

@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json

@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/metadata.json",

src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json

@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/7aa3344bf84c42467f43b3a927e1c7d13642bd55/src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
-      "commit": "7aa3344bf84c42467f43b3a927e1c7d13642bd55",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/9e6a51cab505a8ae17575dc83ea53fc0d3b8b027/src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
+      "commit": "9e6a51cab505a8ae17575dc83ea53fc0d3b8b027",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/metadata.json",