wip v28

Author: geraldarthurdavis

BITCODE_SPEC_V28.md

@@ -474,7 +474,10 @@ Read/Fit result review remains fail-closed:
   incremental telemetry artifact lines while the detached sandbox command is
   still running. Browser-only Network inspection is not sufficient; the
   operator-visible stream must show phase, agent, generation/tool, parsed
-  output, and failsafe context as first-class Read/Fit run evidence.
+  output, and failsafe context as first-class Read/Fit run evidence. Terminal
+  must render these events through the canonical Bitcode execution stream panel
+  shape so Read/Fit live runs are visually consistent with persisted execution
+  activity and retain expandable raw metadata per line.
 - the harness must enforce an internal runtime budget before the caller's host
   timeout, producing a `blocked_readiness` artifact rather than allowing the
   host to terminate without evidence.

BITCODE_V28_QA.md

@@ -929,7 +929,10 @@ First-run execution boundary:
 - The Terminal live-run panel must show the active Read id, Deposit id,
   source commit, sandbox id when available, pipeline run id when available, and
   incremental `telemetry.jsonl` line summaries while the sandbox command is
-  still running. A state visible only in browser Network logs fails V28 QA.
+  still running. It must use the canonical Bitcode execution stream component,
+  with run identifiers in the header metadata and each host/telemetry event as
+  an expandable line carrying raw payload metadata. A state visible only in
+  browser Network logs fails V28 QA.
 - Depository vector recall uses `text-embedding-3-small` by default with
   `encoding_format='float'`, `dimensions=1536`, Supabase
   `deliverable_vectors.embedding vector(1536)`, `ivfflat`,
@@ -1239,7 +1242,9 @@ After either harness run:
    and phase/agent/step/failsafe/generation correlation.
 4. Inspect the Terminal live-run panel for the same run id and sandbox id. The
    panel must include host lifecycle events and incremental telemetry artifact
-   event summaries, not only the final harness completion payload.
+   event summaries, not only the final harness completion payload. Read/Fit
+   live harness streams should render in the same execution stream UX used by
+   persisted Bitcode activity.
 5. Run saved query
    `supabase/queries/v28_qa_terminal_07_pipeline_harness_after_fit.sql`.
 6. Rerun `v28_qa_terminal_06_read_fit_quality_after_read` and
@@ -1324,6 +1329,30 @@ Observed staging-testnet harness evidence on 2026-05-17:
   harness changes are deployed in the deposited source revision. Only a clean
   no-overlay run may write and read back ledger settlement rows.
 
+Later staging-testnet evidence on 2026-05-17 after the no-overlay deployment:
+
+- A new Deposit/Read/Fit harness run against
+  `engineeredsoftware/ENGI@main:ee1481634c985afbc349f8d8b837cd1c43a254ac`
+  reached real model-backed setup execution. Umbrella pipeline run
+  `44a05fd7-f337-42cf-ad3c-d4a607d54a2b` and deliverable pipeline run
+  `b1b04a2d-0376-4200-b08c-7936076f2566` failed before candidate recall or
+  AssetPack synthesis.
+- Database readback showed real execution progress: 2 setup phase delegation
+  rows, 16 agent-step rows, 72 model-generation rows, 1634 deliverable pipeline
+  events, and 0 tool-execution rows. This confirms real inference and telemetry
+  persistence were active, but the run did not reach depository search,
+  synthesis, Finish, shipping, or ledger settlement.
+- Failure root cause was a setup risk-admission output contract mismatch:
+  `asset-pack-danger-wall-agent.ts` read `finalAssessment.safe` directly from
+  the PTRR envelope, while the full PTRR path returns the typed risk result
+  under `output`/`finalOutput`. The fix is to normalize the risk-admission
+  envelope before safety evaluation and fail closed if the typed final
+  assessment is absent.
+- Terminal live-run visibility must now render harness lifecycle events and
+  `telemetry.jsonl` events through the canonical execution stream component so
+  future failures show the same phase/agent/step/generation metadata operators
+  already use for persisted Bitcode activity.
+
 ## 2026-05-13 Staging Deployment Readiness Gate
 
 Purpose:

packages/generic-agents/read-comprehension/README.md

@@ -1,6 +1,6 @@
 # Bitcode Read Comprehension Agent
 
-## V26 Boundary
+## Active Boundary
 
 `@bitcode/generic-agents-read-comprehension` is the PTRR agent owner for Bitcode setup-phase Read comprehension.
 It composes the callable tools from `@bitcode/generic-tools-read-comprehension`; it does not redefine those tool contracts.
@@ -22,7 +22,7 @@ Its canonical fifth-gate role is `setup` / `pre-danger-wall`: synthesize a revie
 - Plans how to understand the expressed Read, attachments, repository context, AssetPack expectation, proof reads, delivery-mechanism boundary, and source-to-shares service questions.
 - Uses Read-comprehension tools to analyze semantics, extract requirements, identify constraints, generate satisfaction criteria, validate the model, and estimate implementation complexity.
 - Refines ambiguity before risk admission rather than allowing danger-wall to infer canonical Read meaning.
-- Emits active Bitcode fields: `read`, `writtenAssetTypes`, `assetPackContext`, `deliveryMechanismBoundaries`, `sourceToSharesServiceQuestions`, `commercialAccountability`, and `readSatisfactionCriteria`.
+- Emits active Bitcode fields: `read`, `writtenAssetTypes`, `assetPackContext`, `deliveryMechanismBoundaries`, `sourceToSharesServiceQuestions`, `serviceAccountability`, and `readSatisfactionCriteria`.
 
 ## Verification
 

packages/generic-agents/read-comprehension/src/index.ts

@@ -85,10 +85,10 @@ export const BitcodeSetupReadComprehensionResultSchema = z.object({
   asset_pack_context: z.any().optional(),
   delivery_mechanism_boundaries: z.array(z.string()).default([]),
   source_to_shares_service_questions: z.record(z.string()).optional(),
-  commercial_accountability: z.object({
+  service_accountability: z.object({
     provider: z.string().optional(),
-    customer_outcome: z.string().optional(),
-    market_infrastructure_standard: z.string().optional()
+    reader_outcome: z.string().optional(),
+    infrastructure_standard: z.string().optional()
   }).optional(),
   comprehension: z.object({
     intent: z.string().optional(),

packages/generic-tools/read-comprehension/README.md

@@ -1,6 +1,6 @@
 # Read Comprehension Tools
 
-## V26 Posture
+## Active Posture
 
 `@bitcode/generic-tools-read-comprehension` is the canonical generic-tool package for Bitcode Read comprehension.
 The package path and package name are canonical Bitcode tool owners.
@@ -40,7 +40,7 @@ Expected output includes:
 - `semantic_analysis.scope_boundaries`
 - `written_asset_expectations`
 - `source_to_shares_service_questions`
-- `commercial_accountability`
+- `service_accountability`
 - `asset_pack_context`
 - `delivery_mechanism_boundaries`
 
@@ -55,7 +55,7 @@ Expected output includes:
 - `functional_requirements`
 - `non_functional_requirements`
 - `proof_requirements`
-- `commercial_service_requirements`
+- `service_requirements`
 - `interface_requirements`
 - `written_asset_requirement_map`
 - `extraction_metadata`
@@ -85,7 +85,7 @@ Expected output includes:
 - `performance_criteria`
 - `quality_criteria`
 - `proof_criteria`
-- `commercial_accountability_criteria`
+- `service_accountability_criteria`
 - `persistence_criteria`
 - `interface_criteria`
 - `blocking_criteria`
@@ -100,7 +100,7 @@ Expected output includes:
 - `validation_results`
 - `validation_details`
 - `terminology_findings`
-- `commercial_accountability_findings`
+- `service_accountability_findings`
 - `read_comprehension`
 - `written_asset_coherence`
 - `proof_coverage`

packages/generic-tools/read-comprehension/src/read-comprehension-primitives.ts

@@ -62,7 +62,7 @@ export async function analyzeReadSemantics({
     },
     written_asset_expectations: ['source-bearing written asset', 'verification evidence'],
     source_to_shares_service_questions: buildSourceToSharesServiceQuestions(expressedRead),
-    commercial_accountability: {
+    service_accountability: {
       provider: 'Advanced Engineered Software, Inc.',
       customer_outcome: 'source-to-shares system use',
       market_infrastructure_standard: 'auditable engineering-knowledge market infrastructure'
@@ -133,7 +133,7 @@ export async function extractReadRequirements({
     business_requirements: [],
     technical_requirements: [],
     proof_requirements: ['Spec/proof witnesses updated when semantics change'],
-    commercial_service_requirements: [
+    service_requirements: [
       'Read answer explains why/how/when/where/what/who the work serves source-to-shares',
       'Proof evidence supports customer-facing audit and market-infrastructure acceptance'
     ],
@@ -264,7 +264,7 @@ export async function generateReadSatisfactionCriteria({
       }
     ],
     proof_criteria: ['Generated V26 proof artifacts include the changed prompt surface'],
-    commercial_accountability_criteria: [
+    service_accountability_criteria: [
       'Critical source-to-shares service questions are answered before fit search or settlement',
       'The answer is reviewable by the customer/operator using canonical Bitcode terminology'
     ],
@@ -321,7 +321,7 @@ export async function validateReadComprehension({
       written_asset_coherence: 0.9,
       proof_coverage: 0.86
     },
-    commercial_accountability_findings: [
+    service_accountability_findings: [
       'Read comprehension must explain why/how/when/where/what/who this code serves Bitcode source-to-shares'
     ],
     terminology_findings: [
@@ -366,7 +366,7 @@ export async function analyzeReadSatisfactionImplementationComplexity({
       testing_complexity: 'medium' as const,
       proof_complexity: 'medium' as const,
       delivery_mechanism_complexity: 'low' as const,
-      commercial_accountability_complexity: 'medium' as const
+      service_accountability_complexity: 'medium' as const
     },
     risk_analysis: {
       technical_risks: ['Package boundary drift', 'Prompt runtime carry-through drift'],

packages/pipelines/asset-pack/src/__tests__/depository-search.test.ts

@@ -59,6 +59,11 @@ function asset(overrides: Partial<DepositoryAsset> = {}): DepositoryAsset {
     githubBoundary: { sourceProvider: 'github', sourceRepo: 'engineeredsoftware/ENGI' },
     assetMeasurement: { targetKindCount: 5 },
     measurementProvenance: [{ stage: 'deposit-measurement' }],
+    verificationEvidence: {
+      proofRoot: 'sha256:test-proof-root',
+      measurementRoot: 'sha256:test-measurement-root',
+      reconciliationReadbackRoot: 'sha256:test-reconciliation-readback-root',
+    },
     ...overrides,
   };
 }
@@ -140,6 +145,25 @@ describe('AssetPack depository search', () => {
     );
   });
 
+  it('blocks readiness when a read requires proof and reconciliation readback roots that are not present', async () => {
+    const result = await searchDepositoryAssetSpace({
+      read,
+      assets: [
+        asset({
+          verificationEvidence: null,
+          hasWalletOrAttestationProof: true,
+          hasAssetMeasurementEvidence: true,
+        }),
+      ],
+    });
+
+    expect(result.resultState).toBe('blocked_readiness');
+    expect(result.selectedCandidates[0].verification.warnings).toEqual(
+      expect.arrayContaining(['proof_root_readback_missing', 'reconciliation_readback_missing'])
+    );
+    expect(result.selectedCandidates[0].useTier).toBe('context-only');
+  });
+
   it('returns no-worthy-fit for unrelated reads instead of relying on repository match alone', async () => {
     const result = await searchDepositoryAssetSpace({
       read: {
@@ -240,10 +264,16 @@ describe('AssetPack depository search', () => {
       proofEvidence: {
         hasWalletOrAttestationProof: true,
         signingSurfacePresent: true,
+        proofRoot: 'sha256:test-proof-root',
       },
       measurementEvidence: {
         hasAssetMeasurementEvidence: true,
         assetMeasurementPresent: true,
+        measurementRoot: 'sha256:test-measurement-root',
+      },
+      readbackEvidence: {
+        reconciliationReadbackPresent: true,
+        reconciliationReadbackRoot: 'sha256:test-reconciliation-readback-root',
       },
     });
     expect(findStored(exec, 'fit', 'selectionTrace')?.selectedCandidates[0].selectedUnits[0]).toMatchObject({

packages/pipelines/asset-pack/src/__tests__/setup-agents.test.ts

@@ -2,7 +2,9 @@
 import cloneRepositoryAgent from '../agents/setup/asset-pack-clone-vcs-repository-agent';
 import setupPlanAgent from '../agents/setup/asset-pack-setup-plan-agent';
 import runComprehendReadAgent from '../agents/setup/asset-pack-comprehend-read-agent';
-import dangerWallAgent from '../agents/setup/asset-pack-danger-wall-agent';
+import dangerWallAgent, {
+  normalizeRiskAdmissionResult,
+} from '../agents/setup/asset-pack-danger-wall-agent';
 
 function executionStub() {
   const stores: Array<{ namespace: string; key: string; value: unknown }> = [];
@@ -137,4 +139,66 @@ describe('AssetPack setup agents', () => {
       ])
     );
   });
+
+  it('normalizes the full PTRR risk admission envelope before checking setup safety', () => {
+    const result = normalizeRiskAdmissionResult({
+      context: {},
+      output: {
+        finalAssessment: {
+          safe: true,
+          maxSeverity: 'none',
+          confidence: 0.91,
+          verdict: {
+            approved: true,
+            reason: 'Full risk admission passed.',
+            flags: [],
+            recommendations: [],
+          },
+          auditTrail: [
+            {
+              check: 'read boundary',
+              result: true,
+              details: ['source-bound read'],
+              severity: 'none',
+            },
+          ],
+        },
+        riskInsights: {
+          riskProfile: 'bounded',
+          threatLevel: 'minimal',
+          riskRecommendations: [],
+          proofObligations: [],
+          admissionBoundary: 'setup',
+        },
+        readAlignment: {
+          alignmentScore: 0.94,
+          readSafeToMeasure: true,
+          assetPackSafeToSynthesize: true,
+          deliveryMechanismSafeToAttempt: true,
+        },
+        recommendations: ['continue'],
+        success: true,
+        validationMessage: 'passed',
+      },
+      finalOutput: {},
+    });
+
+    expect(result.finalAssessment.safe).toBe(true);
+    expect(result.finalAssessment.maxSeverity).toBe('none');
+    expect(result.readAlignment.assetPackSafeToSynthesize).toBe(true);
+  });
+
+  it('fails closed when risk admission returns an untyped output shape', () => {
+    const result = normalizeRiskAdmissionResult({
+      step: 'Retry',
+      agent: 'bitcode-read-risk-admission',
+      phase: 'setup',
+    });
+
+    expect(result.finalAssessment.safe).toBe(false);
+    expect(result.finalAssessment.maxSeverity).toBe('high');
+    expect(result.finalAssessment.verdict.flags).toContain(
+      'risk-admission-output-missing-final-assessment',
+    );
+  });
 });

packages/pipelines/asset-pack/src/agents/setup/asset-pack-comprehend-read-agent.ts

@@ -139,10 +139,10 @@ function buildDeterministicReadComprehension(input: any, execution: any) {
       'BTC fee handling is a settlement/finality concern and cannot be attributed to the depositor without recorded authorization.'
     ],
     source_to_shares_service_questions: {},
-    commercial_accountability: {
+    service_accountability: {
       provider: 'Bitcode pipeline harness',
-      customer_outcome: 'A high-quality, proof-bearing AssetPack fit or explicit blocked/no-worthy-fit evidence.',
-      market_infrastructure_standard: 'Source-bound read satisfaction with auditable telemetry and ledger synchronization.'
+      reader_outcome: 'A high-quality, proof-bearing AssetPack fit or explicit blocked/no-worthy-fit evidence.',
+      infrastructure_standard: 'Source-bound read satisfaction with auditable telemetry and ledger synchronization.'
     },
     comprehension: {
       intent: 'Synthesize a Read-satisfying AssetPack from deposited repository evidence.',