Merge branch 'main' into docs/discovery-plane-operator-guide-finalize

Author: JSONbored

.github/workflows/ci.yml

@@ -1028,7 +1028,21 @@ jobs:
             # ~2,900-file suite is 0% covered, polluting Codecov's project trend on every scoped PR. Real
             # coverage for files actually exercised is unaffected either way.
             SCOPE_ARGS+=(--changed=origin/main --coverage.all=false)
-            npm run test:coverage -- --maxWorkers=4 --shard=${{ matrix.shard }}/3 --reporter=default --reporter=blob --reporter=junit --outputFile.blob=blob-report/report-${{ matrix.shard }}.blob --outputFile.junit=reports/junit/vitest.xml "${EXCLUDE_ARGS[@]}" "${SCOPE_ARGS[@]}"
+            # A scoped PR touching zero files any test imports (a pure docs/comment-only diff under a
+            # scoped path, e.g. packages/loopover-miner/docs/**) is a real, valid outcome here -- vitest
+            # itself prints "No test files found, exiting with code 0" and exits clean, writing no
+            # coverage/lcov.info at all (nothing was instrumented). Confirmed live: PR #8165, a pure
+            # packages/loopover-miner/docs + apps/loopover-ui/content/docs mdx change, matched zero test
+            # files and failed "Verify coverage report exists" even though nothing was actually broken.
+            # Capture vitest's own stdout (the same "No test files found" string this file's own comment
+            # above already documents relying on) rather than re-deriving the zero-files case from
+            # scratch, and record it as a step output so "Verify coverage report exists" below can tell
+            # this apart from a real crash that should still fail loudly.
+            set -o pipefail
+            npm run test:coverage -- --maxWorkers=4 --shard=${{ matrix.shard }}/3 --reporter=default --reporter=blob --reporter=junit --outputFile.blob=blob-report/report-${{ matrix.shard }}.blob --outputFile.junit=reports/junit/vitest.xml "${EXCLUDE_ARGS[@]}" "${SCOPE_ARGS[@]}" 2>&1 | tee vitest-scoped-output.log
+            if grep -q "No test files found" vitest-scoped-output.log; then
+              echo "no_tests_matched=true" >> "$GITHUB_OUTPUT"
+            fi
           else
             # Duration-aware sharding (#ci-duration-aware-sharding), full-suite case only: vitest's own
             # --shard splits by file COUNT alone, no duration awareness -- confirmed via real per-shard CI
@@ -1060,7 +1074,12 @@ jobs:
           echo "Coverage itself is gated by Codecov on changed lines (codecov/patch), computed from all shards' merged lcov."
           echo "Reproduce locally with: 'npm run test:coverage' (unsharded, runs the whole suite)."
       - name: Verify coverage report exists
-        if: ${{ success() }}
+        # Skipped when the scoped-selection branch above matched zero test files (steps.coverage.outputs.
+        # no_tests_matched) -- a legitimately test-free diff (e.g. docs-only under a scoped path) writes no
+        # coverage/lcov.info at all, and that is not a failure. A real failure (tests ran and either failed
+        # or the coverage step crashed before writing output) still fails this job via `success()` below,
+        # same as before.
+        if: ${{ success() && steps.coverage.outputs.no_tests_matched != 'true' }}
         run: |
           if [ ! -s coverage/lcov.info ]; then
             echo "::error title=Coverage::coverage/lcov.info is missing or empty"

.loopover.yml.example

@@ -329,6 +329,16 @@ gate:
   # value.
   contentLaneDeliverable: off
 
+  # Backtest-regression gate (#8105). Governs what a REGRESSED verdict from the
+  # pre-merge backtest (threshold #8138 / logic #8139) does: advisory (default)
+  # renders the comparison in the review comment but never blocks; block
+  # escalates it into a backtest_regression hard blocker; off silences the
+  # backtest advisory entirely. Flip to block only once the persisted track
+  # record (scripts/backtest-track-record.ts, #8140) supports it.
+  # off | advisory | block. Default: advisory. DB-backed (dashboard-settable
+  # too); this overrides the stored value.
+  backtestRegression: advisory
+
   # Gate-check dry-run. When true, the posted check conclusion remains the real
   # non-enforcing verdict while comments/check text may also show the would-be
   # stricter verdict for AI-review blocker mode. It does not disable downstream

apps/loopover-ui/content/docs/backtest-calibration.mdx

@@ -15,9 +15,10 @@ threshold or rewrites detection logic gets scored against the real recorded hist
 targets, the same raw inputs — instead of being eyeballed.
 
 <Callout variant="note">
-  Everything on this page is advisory-only today. A backtest verdict — even a regression — never
-  blocks a merge. Whether a REGRESSED verdict should ever gate merges is a tracked, deliberately
-  separate decision that waits for real production track-record data.
+  Everything on this page is advisory-only by default. A REGRESSED verdict blocks a merge only if a
+  repo explicitly opts in via `backtestRegressionGateMode: block` (`.loopover.yml
+  gate.backtestRegression`) — the shipped default is `advisory`, and the flip is meant to wait for
+  real production track-record data.
 </Callout>
 
 ## The corpus

apps/loopover-ui/content/docs/tuning.mdx

@@ -259,6 +259,12 @@ forward and wins whenever both are set for the same field.
   `advisory` renders the assessment in the review comment without blocking;
   `block` additionally lets a confidence-floor-passing "unaddressed" verdict
   become a blocker.
+- `gate.backtestRegression` — governs what a REGRESSED verdict from the
+  [pre-merge backtest](/docs/backtest-calibration) does. Default `advisory`:
+  the comparison renders in the review comment but never blocks. `block`
+  escalates a REGRESSED verdict into a hard blocker — flip it only once the
+  persisted track record supports gating. `off` silences the backtest advisory
+  entirely.
 - `gate.contentLaneDeliverable` — only meaningful for a repo with a registry
   content-lane spec configured (`contentLane`); a no-op otherwise. Fully
   deterministic (a text/path match, no AI call): when the PR's primary linked
@@ -381,36 +387,42 @@ testExpectations:
 linkedIssuePolicy: preferred
 
 # Gate policy — checkMode is set explicitly (not the legacy, ambiguous "enabled" alias)
+
 gate:
-  checkMode: visible
-  pack: gittensor
-  duplicates: block
-  linkedIssue: advisory
-  readiness:
-    mode: advisory
-    minScore: 70
-  slop:
-    mode: block
-    minScore: 60
-    aiAdvisory: true
-  mergeReadiness: advisory
-  manifestPolicy: block
-  aiReview:
-    mode: advisory
-    byok: true
-    provider: anthropic
-    model: claude-3-5-sonnet-latest
+checkMode: visible
+pack: gittensor
+duplicates: block
+linkedIssue: advisory
+readiness:
+mode: advisory
+minScore: 70
+slop:
+mode: block
+minScore: 60
+aiAdvisory: true
+mergeReadiness: advisory
+manifestPolicy: block
+aiReview:
+mode: advisory
+byok: true
+provider: anthropic
+model: claude-3-5-sonnet-latest
 
 # Generic settings overrides -- commentMode/checkRunMode/checkRunDetailLevel/badgeEnabled are
+
 # config-as-code only (no DB column or dashboard toggle); this file is their sole source.
+
 settings:
-  commentMode: detected_contributors_only
-  checkRunMode: enabled
-  checkRunDetailLevel: standard
-  badgeEnabled: true
-  # Optional path holds. Omitted or [] means no path guardrails.
-  # hardGuardrailGlobs:
-  #   - "src/selfhost/**"`}
+commentMode: detected_contributors_only
+checkRunMode: enabled
+checkRunDetailLevel: standard
+badgeEnabled: true
+
+# Optional path holds. Omitted or [] means no path guardrails.
+
+# hardGuardrailGlobs:
+
+# - "src/selfhost/\*\*"`}
 
 />
 

apps/loopover-ui/public/openapi.json

@@ -9761,6 +9761,14 @@
               "advisory",
               "block"
             ]
+          },
+          "backtestRegressionGateMode": {
+            "type": "string",
+            "enum": [
+              "off",
+              "advisory",
+              "block"
+            ]
           }
         },
         "required": [
@@ -9782,6 +9790,7 @@
           "selfAuthoredLinkedIssueGateMode",
           "linkedIssueSatisfactionGateMode",
           "contentLaneDeliverableGateMode",
+          "backtestRegressionGateMode",
           "slopAiAdvisory",
           "aiReviewMode",
           "aiReviewByok",
@@ -10492,6 +10501,14 @@
                   "advisory",
                   "block"
                 ]
+              },
+              "backtestRegressionGateMode": {
+                "type": "string",
+                "enum": [
+                  "off",
+                  "advisory",
+                  "block"
+                ]
               }
             },
             "required": [
@@ -10513,6 +10530,7 @@
               "selfAuthoredLinkedIssueGateMode",
               "linkedIssueSatisfactionGateMode",
               "contentLaneDeliverableGateMode",
+              "backtestRegressionGateMode",
               "autoLabelEnabled",
               "typeLabelsEnabled",
               "gittensorLabel",

config/examples/loopover.full.yml

@@ -343,6 +343,16 @@ gate:
   # value.
   contentLaneDeliverable: off
 
+  # Backtest-regression gate (#8105). Governs what a REGRESSED verdict from the
+  # pre-merge backtest (threshold #8138 / logic #8139) does: advisory (default)
+  # renders the comparison in the review comment but never blocks; block
+  # escalates it into a backtest_regression hard blocker; off silences the
+  # backtest advisory entirely. Flip to block only once the persisted track
+  # record (scripts/backtest-track-record.ts, #8140) supports it.
+  # off | advisory | block. Default: advisory. DB-backed (dashboard-settable
+  # too); this overrides the stored value.
+  backtestRegression: advisory
+
   # Gate-check dry-run. When true, the posted check conclusion remains the real
   # non-enforcing verdict while comments/check text may also show the would-be
   # stricter verdict for AI-review blocker mode. It does not disable downstream

migrations/0174_gate_backtest_regression.sql

@@ -0,0 +1,6 @@
+-- Backtest-regression gate (#8105, epic #8082): advisory by default -- byte-identical behavior for every
+-- existing row (the shipped #8138/#8142 comment-only advisory). block escalates a REGRESSED pre-merge
+-- backtest verdict into a backtest_regression hard blocker; off silences the backtest advisory entirely.
+-- The flip to block is deliberately a config change made only once #8140's persisted track record supports
+-- it -- see #8105's own do-not-gate-before-data boundary.
+ALTER TABLE repository_settings ADD COLUMN backtest_regression_gate_mode TEXT NOT NULL DEFAULT 'advisory';

packages/loopover-engine/src/advisory/gate-advisory.ts

@@ -99,6 +99,10 @@ export type GateCheckPolicy = {
    *  on this mode; see runContentLaneDeliverableCheckForAdvisory, src/queue/processors.ts on the host side),
    *  so this branch only matters once a repo has explicitly opted into `advisory`/`block`. */
   contentLaneDeliverableGateMode?: GateRuleMode | undefined;
+  /** Backtest-regression gate (#8105): `block` turns a `backtest_regression` finding into a hard blocker.
+   *  Default `advisory` — the finding only exists in block mode (the host resolver pushes it there), so
+   *  this branch is defense-in-depth, mirroring content_lane_deliverable_missing's above. */
+  backtestRegressionGateMode?: GateRuleMode | undefined;
   /** CLA / license-compatibility gate (#2564). When `block`, a `cla_consent_missing` finding — raised when
    *  neither configured detection method (a consent phrase in the PR body, or a named CLA-bot check-run
    *  conclusion) confirms consent — becomes a hard blocker. `off` (default) = no finding at all; `advisory` =
@@ -636,6 +640,9 @@ function isConfiguredGateBlocker(finding: AdvisoryFinding, policy: GateCheckPoli
   // opts in with `block`. Fully deterministic (no AI judgment involved), so it is exempt from the
   // close-precision circuit breaker on the host side.
   if (code === "content_lane_deliverable_missing") return gatePolicyBlocks(policy.contentLaneDeliverableGateMode, "off");
+  // Backtest-regression gate (#8105): blocks only under an explicit opt-in; default advisory (the shipped
+  // pre-#8105 behavior). The finding itself only exists in block mode -- see the host-side resolver.
+  if (code === "backtest_regression") return gatePolicyBlocks(policy.backtestRegressionGateMode, "advisory");
   // Lockfile-tamper-risk gate (#2563): blocks only when the maintainer opts in with `block`. Defaults to `off`
   // (the finding is never even produced — see maybeAddLockfileTamperFinding's mode gate in queue/processors.ts),
   // so this branch only matters once a repo has explicitly turned the scan on.

packages/loopover-engine/src/focus-manifest.ts

@@ -165,6 +165,12 @@ export type FocusManifestGateConfig = {
    *  `linkedIssueSatisfaction` immediately above in shape, but is a purely structural check, not an AI
    *  opinion, so it carries none of that feature's AI-budget/confidence-floor machinery. */
   contentLaneDeliverable: GateRuleMode | null;
+  /** `gate.backtestRegression` (#8105, epic #8082): off|advisory|block, advisory by default — governs what a
+   *  REGRESSED pre-merge backtest verdict (#8138 threshold / #8139 logic) does. `advisory` renders the
+   *  comparison but never blocks (the shipped pre-#8105 behavior); `block` escalates it into a
+   *  `backtest_regression` hard blocker; `off` silences the backtest advisory entirely. DB-backed
+   *  (dashboard-settable too); this overrides the stored value like every other `gate:` field. */
+  backtestRegression: GateRuleMode | null;
   dryRun: boolean | null;
   /** `gate.premergeContentRecheck` (#2550): for a PR touching `migrations/**`, re-verify against a live,
    *  freshly-fetched tip of the base branch — unioned with this PR's own new migration filenames — for a
@@ -1303,6 +1309,7 @@ const EMPTY_GATE_CONFIG: FocusManifestGateConfig = {
   selfAuthoredLinkedIssue: null,
   linkedIssueSatisfaction: null,
   contentLaneDeliverable: null,
+  backtestRegression: null,
   dryRun: null,
   premergeContentRecheck: null,
   requireFreshRebaseWindowMinutes: null,
@@ -1788,6 +1795,7 @@ function parseGateConfig(value: JsonValue | undefined, warnings: string[]): Focu
     selfAuthoredLinkedIssue: normalizeOptionalGateMode(record.selfAuthoredLinkedIssue, "gate.selfAuthoredLinkedIssue", warnings),
     linkedIssueSatisfaction: normalizeOptionalGateMode(record.linkedIssueSatisfaction, "gate.linkedIssueSatisfaction", warnings),
     contentLaneDeliverable: normalizeOptionalGateMode(record.contentLaneDeliverable, "gate.contentLaneDeliverable", warnings),
+    backtestRegression: normalizeOptionalGateMode(record.backtestRegression, "gate.backtestRegression", warnings),
     dryRun: normalizeOptionalBoolean(record.dryRun, "gate.dryRun", warnings),
     premergeContentRecheck: normalizeOptionalBoolean(record.premergeContentRecheck, "gate.premergeContentRecheck", warnings),
     requireFreshRebaseWindowMinutes: normalizeOptionalPositiveInteger(record.requireFreshRebaseWindow, "gate.requireFreshRebaseWindow", warnings),
@@ -1845,6 +1853,7 @@ function parseGateConfig(value: JsonValue | undefined, warnings: string[]): Focu
     gate.selfAuthoredLinkedIssue !== null ||
     gate.linkedIssueSatisfaction !== null ||
     gate.contentLaneDeliverable !== null ||
+    gate.backtestRegression !== null ||
     gate.dryRun !== null ||
     gate.premergeContentRecheck !== null ||
     gate.requireFreshRebaseWindowMinutes !== null ||

scripts/check-docs-drift.ts

@@ -240,6 +240,7 @@ export const GATE_MODE_MANIFEST: GateModeManifestRow[] = [
   { field: "selfAuthoredLinkedIssueGateMode", aliases: ["selfAuthoredLinkedIssueGateMode", "gate.selfAuthoredLinkedIssue"], pages: ["how-reviews-work.mdx", "tuning.mdx", "github-app.mdx"] },
   { field: "linkedIssueSatisfactionGateMode", aliases: ["linkedIssueSatisfactionGateMode", "gate.linkedIssueSatisfaction"], pages: ["how-reviews-work.mdx", "tuning.mdx", "github-app.mdx"] },
   { field: "contentLaneDeliverableGateMode", aliases: ["contentLaneDeliverableGateMode", "gate.contentLaneDeliverable"], pages: ["how-reviews-work.mdx", "tuning.mdx", "github-app.mdx"] },
+  { field: "backtestRegressionGateMode", aliases: ["backtestRegressionGateMode", "gate.backtestRegression"], pages: ["backtest-calibration.mdx", "tuning.mdx"] },
   { field: "moderationGateMode", aliases: ["moderationGateMode", "settings.moderationGateMode"], pages: ["how-reviews-work.mdx", "tuning.mdx", "github-app.mdx"] },
 ];