docs: link calibrated native candidate campaign

Co-Authored-By: Paperclip <noreply@paperclip.ing>

doc/plans/2026-10-02-native-completion-calibrated-manifest.json

@@ -266,9 +266,15 @@
   "fixtureIdentityVerified": true,
   "onlyDeclaredElevenNativeSourcesDiffer": true,
   "fixedMasterContextVerified": true,
-  "providerCallsSoFar": 0,
-  "liveStatus": "admitted_not_dispatched",
-  "workflowRuns": {},
+  "providerCallsSoFar": null,
+  "liveStatus": "candidate_dispatched_baseline_approval_pending",
+  "workflowRuns": {
+    "candidate": {
+      "id": 37098728980,
+      "url": "https://github.com/paperclipai/paperclip/actions/runs/37098728980",
+      "createdAt": "2026-10-03T05:06:09Z"
+    }
+  },
   "preservedZeroProviderCohorts": [
     {
       "candidate": "e18c2cf9e96a4d31acb6d03ce918dfe223107d3f",
@@ -292,5 +298,6 @@
       "providerCalls": 0,
       "classification": "pre_provider_calibration_mock"
     }
-  ]
+  ],
+  "workflowSourceSha": "215586d127e97c9301d86e769a39a15c13298ca2"
 }

doc/plans/2026-10-02-native-completion-master-qualification.md

@@ -6,7 +6,7 @@ The [immutable calibrated manifest](2026-10-02-native-completion-calibrated-mani
 
 Only the previously reviewed eleven native implementation/unit-test files differ between variants. All five setup repair files are identical: a controlled mock receipt SHA, first-line framed provenance parsing plus positive/negative calibration, and matching immutable parent declarations. Production default/manual/shared-prompt/operational-skill bytes still match master `59c07ede`. The verifier's exact-source rejection and all behavioral oracles remain strict.
 
-Both previous twelve-cell hosted cohorts stopped before providers and remain preserved below. [The twelve-retained-log replay](2026-10-02-native-completion-calibrated-admission/retained-frame-replay.json) validates exact hashes/framed proofs with no change to original failing verdicts. No model behavior reroll has occurred. Actual matched dispatch is the next step; this qualification does not measure fixed native prompt removal or general coding quality.
+Both previous twelve-cell hosted cohorts stopped before providers and remain preserved below. [The twelve-retained-log replay](2026-10-02-native-completion-calibrated-admission/retained-frame-replay.json) validates exact hashes/framed proofs with no change to original failing verdicts. No model behavior reroll has occurred. [Candidate campaign 37098728980](https://github.com/paperclipai/paperclip/actions/runs/37098728980) is dispatched against immutable `d6e59e471`; the trusted workflow source is `215586d127e97c9301d86e769a39a15c13298ca2`. No graded cells are available yet. The historical dispatch is pending normal approval review: its initial request was rejected based on an earlier no-additional-runs instruction, and the parent task is submitting the same bounded request with the current direct human authorization. No duplicate candidate or baseline has been launched. Current campaign usage/cost is not yet available. This qualification does not measure fixed native prompt removal or general coding quality.
 
 ## Preserved prior qualification stages