@openwop/openwop-conformance 1.162.0 → 2.0.0-rc.0

package/src/scenarios/agent-eval-suite-shape.test.ts

@@ -33,9 +33,7 @@ import { join } from 'node:path';
 import Ajv2020 from 'ajv/dist/2020.js';
 import addFormats from 'ajv-formats';
 import { SCHEMAS_DIR } from '../lib/paths.js';
-
-/** Server-free assertion-message helper (mirrors driver.describe's "spec — requirement" shape without requiring OPENWOP_BASE_URL). */
-const why = (specRef: string, requirement: string): string => `${specRef} — ${requirement}`;
+import { req } from '../lib/requirement-ids.js';
 
 function loadSchema(name: string): Record<string, unknown> {
   return JSON.parse(readFileSync(join(SCHEMAS_DIR, name), 'utf8')) as Record<string, unknown>;
@@ -48,12 +46,12 @@ describe('agent-eval-suite-shape: capability advertisement (RFC 0081, server-fre
     const evalSuite = agents?.properties?.evalSuite;
     expect(
       evalSuite,
-      why('capabilities.md §agents', 'agents.evalSuite MUST be declared'),
+      req('openwop.it.agent-eval-suite-shape.the-capabilities-schema-declares-agents-evalsuite-with-its-sub-flags', 'capabilities.md §agents', 'agents.evalSuite MUST be declared'),
     ).toBeDefined();
     for (const flag of ['supported', 'modes']) {
       expect(
         evalSuite?.properties?.[flag],
-        why('agent-evaluation.md §Capability advertisement', `agents.evalSuite.${flag} MUST be declared`),
+        req('openwop.it.agent-eval-suite-shape.the-capabilities-schema-declares-agents-evalsuite-with-its-sub-flags', 'agent-evaluation.md §Capability advertisement', `agents.evalSuite.${flag} MUST be declared`),
       ).toBeDefined();
     }
   });
@@ -75,11 +73,11 @@ describe('agent-eval-suite-shape: AgentEvalSuite + EvalSummary schemas (RFC 0081
         { taskId: 'refund-window', input: { q: 'refund policy?' }, expected: { kind: 'golden', match: { strategy: 'contains', value: '30 days' } } },
       ],
     };
-    expect(suite(good), why('RFC 0081 §A', 'a conforming AgentEvalSuite MUST validate')).toBe(true);
+    expect(suite(good), req('openwop.it.agent-eval-suite-shape.agentevalsuite-validates-a-conforming-suite-and-rejects-a-malformed-suiteid-out', 'RFC 0081 §A', 'a conforming AgentEvalSuite MUST validate')).toBe(true);
     // Negative: suiteId must carry the `.evals.` infix.
-    expect(suite({ ...good, suiteId: 'core.openwop.support-resolver' }), why('RFC 0081 §A', 'a suiteId without the `.evals.` infix MUST be rejected')).toBe(false);
+    expect(suite({ ...good, suiteId: 'core.openwop.support-resolver' }), req('openwop.it.agent-eval-suite-shape.agentevalsuite-validates-a-conforming-suite-and-rejects-a-malformed-suiteid-out', 'RFC 0081 §A', 'a suiteId without the `.evals.` infix MUST be rejected')).toBe(false);
     // Negative: passScore out of 0..1.
-    expect(suite({ ...good, thresholds: { passScore: 1.5 } }), why('RFC 0081 §A', 'thresholds.passScore > 1 MUST be rejected')).toBe(false);
+    expect(suite({ ...good, thresholds: { passScore: 1.5 } }), req('openwop.it.agent-eval-suite-shape.agentevalsuite-validates-a-conforming-suite-and-rejects-a-malformed-suiteid-out', 'RFC 0081 §A', 'thresholds.passScore > 1 MUST be rejected')).toBe(false);
   });
 
   it('EvalSummary validates a conforming scorecard and rejects an out-of-range score', () => {
@@ -92,8 +90,8 @@ describe('agent-eval-suite-shape: AgentEvalSuite + EvalSummary schemas (RFC 0081
       passedCount: 2,
       tasks: [{ taskId: 'refund-window', score: 0.9, passed: true, safetyFindings: [{ kind: 'jailbreak', severity: 'low' }] }],
     };
-    expect(summary(good), why('RFC 0081 §C', 'a conforming EvalSummary MUST validate')).toBe(true);
-    expect(summary({ ...good, aggregateScore: 1.4 }), why('RFC 0081 §C', 'aggregateScore > 1 MUST be rejected')).toBe(false);
+    expect(summary(good), req('openwop.it.agent-eval-suite-shape.evalsummary-validates-a-conforming-scorecard-and-rejects-an-out-of-range-score', 'RFC 0081 §C', 'a conforming EvalSummary MUST validate')).toBe(true);
+    expect(summary({ ...good, aggregateScore: 1.4 }), req('openwop.it.agent-eval-suite-shape.evalsummary-validates-a-conforming-scorecard-and-rejects-an-out-of-range-score', 'RFC 0081 §C', 'aggregateScore > 1 MUST be rejected')).toBe(false);
   });
 
   it('EvalSummary is content-free — a task-output body and a safety-finding excerpt are rejected (eval-summary-no-content-leak)', () => {
@@ -101,12 +99,12 @@ describe('agent-eval-suite-shape: AgentEvalSuite + EvalSummary schemas (RFC 0081
     // Negative: a per-task entry carrying the output body.
     expect(
       summary({ ...base, tasks: [{ taskId: 't1', score: 0.5, passed: false, taskOutput: 'the model said …' }] }),
-      why('SECURITY invariant eval-summary-no-content-leak', 'an EvalSummary task entry MUST NOT carry an output body'),
+      req('openwop.it.agent-eval-suite-shape.evalsummary-is-content-free-a-task-output-body-and-a-safety-finding-excerpt-are', 'SECURITY invariant eval-summary-no-content-leak', 'an EvalSummary task entry MUST NOT carry an output body'),
     ).toBe(false);
     // Negative: a safety finding carrying excerpted content rather than a {kind, severity} descriptor.
     expect(
       summary({ ...base, tasks: [{ taskId: 't1', score: 0.5, passed: false, safetyFindings: [{ kind: 'pii-leak', severity: 'high', excerpt: 'SSN 123-45-6789' }] }] }),
-      why('SECURITY invariant eval-summary-no-content-leak', 'a safetyFinding MUST NOT carry excerpted content'),
+      req('openwop.it.agent-eval-suite-shape.evalsummary-is-content-free-a-task-output-body-and-a-safety-finding-excerpt-are', 'SECURITY invariant eval-summary-no-content-leak', 'a safetyFinding MUST NOT carry excerpted content'),
     ).toBe(false);
   });
 });
@@ -122,45 +120,45 @@ describe('agent-eval-suite-shape: eval event payloads (RFC 0081, server-free)',
   const completed = ajv.getSchema('payloads#/$defs/evalCompleted');
 
   it('eval.started validates a content-free start record and requires the suite provenance', () => {
-    expect(started, 'the evalStarted $def MUST exist').toBeTruthy();
+    expect(started, req('openwop.it.agent-eval-suite-shape.eval-started-validates-a-content-free-start-record-and-requires-the-suite-proven', 'RFC 0081', 'the evalStarted $def MUST exist')).toBeTruthy();
     expect(
       started!({ suiteId: 'core.openwop.evals.support-resolver', suiteVersion: '1.0.0', taskCount: 12, modes: ['golden'] }),
-      why('RFC 0081 §C', 'a conforming eval.started payload MUST validate'),
+      req('openwop.it.agent-eval-suite-shape.eval-started-validates-a-content-free-start-record-and-requires-the-suite-proven', 'RFC 0081 §C', 'a conforming eval.started payload MUST validate'),
     ).toBe(true);
     expect(
       started!({ suiteId: 'core.openwop.evals.x' }),
-      why('RFC 0081 §C', 'eval.started without suiteVersion/taskCount/modes MUST be rejected'),
+      req('openwop.it.agent-eval-suite-shape.eval-started-validates-a-content-free-start-record-and-requires-the-suite-proven', 'RFC 0081 §C', 'eval.started without suiteVersion/taskCount/modes MUST be rejected'),
     ).toBe(false);
   });
 
   it('eval.scored validates a content-free per-task score and requires score + passed', () => {
-    expect(scored, 'the evalScored $def MUST exist').toBeTruthy();
+    expect(scored, req('openwop.it.agent-eval-suite-shape.eval-scored-validates-a-content-free-per-task-score-and-requires-score-passed', 'RFC 0081', 'the evalScored $def MUST exist')).toBeTruthy();
     expect(
       scored!({ taskId: 'refund-window', score: 0.9, passed: true, costUsd: 0.012 }),
-      why('RFC 0081 §C', 'a conforming eval.scored payload MUST validate'),
+      req('openwop.it.agent-eval-suite-shape.eval-scored-validates-a-content-free-per-task-score-and-requires-score-passed', 'RFC 0081 §C', 'a conforming eval.scored payload MUST validate'),
     ).toBe(true);
     expect(
       scored!({ taskId: 'refund-window' }),
-      why('RFC 0081 §C', 'eval.scored without score/passed MUST be rejected'),
+      req('openwop.it.agent-eval-suite-shape.eval-scored-validates-a-content-free-per-task-score-and-requires-score-passed', 'RFC 0081 §C', 'eval.scored without score/passed MUST be rejected'),
     ).toBe(false);
   });
 
   it('eval.completed validates a content-free aggregate record', () => {
-    expect(completed, 'the evalCompleted $def MUST exist').toBeTruthy();
+    expect(completed, req('openwop.it.agent-eval-suite-shape.eval-completed-validates-a-content-free-aggregate-record', 'RFC 0081', 'the evalCompleted $def MUST exist')).toBeTruthy();
     expect(
       completed!({ aggregateScore: 0.86, passed: true, taskCount: 12, passedCount: 11, regressionVsBaseline: 0.04 }),
-      why('RFC 0081 §C', 'a conforming eval.completed payload MUST validate'),
+      req('openwop.it.agent-eval-suite-shape.eval-completed-validates-a-content-free-aggregate-record', 'RFC 0081 §C', 'a conforming eval.completed payload MUST validate'),
     ).toBe(true);
     expect(
       completed!({ aggregateScore: 2 }),
-      why('RFC 0081 §C', 'eval.completed with an out-of-range aggregateScore MUST be rejected'),
+      req('openwop.it.agent-eval-suite-shape.eval-completed-validates-a-content-free-aggregate-record', 'RFC 0081 §C', 'eval.completed with an out-of-range aggregateScore MUST be rejected'),
     ).toBe(false);
   });
 
   it('all three eval event names appear in the RunEventType enum', () => {
     const runEvent = loadSchema('run-event.schema.json');
     const enumVals = (runEvent.$defs as Record<string, { enum?: string[] }>).RunEventType?.enum ?? [];
-    expect(enumVals).toContain('eval.started');
+    expect(enumVals, req('openwop.it.agent-eval-suite-shape.all-three-eval-event-names-appear-in-the-runeventtype-enum', 'RFC 0081', 'all three eval event names appear in the RunEventType enum')).toContain('eval.started');
     expect(enumVals).toContain('eval.scored');
     expect(enumVals).toContain('eval.completed');
   });

package/src/scenarios/agent-live-allowlist-enforced.test.ts

@@ -20,10 +20,11 @@
  */
 
 import { describe, it, expect } from 'vitest';
-import { driver } from '../lib/driver.js';
 import { behaviorGate } from '../lib/behavior-gate.js';
 import { readLiveRuntimeCap, invokeLive } from '../lib/liveRuntime.js';
 import { queryTestEvents, isEventLogSeamAvailable, resetTestSeam } from '../lib/event-log-query.js';
+import { req } from '../lib/requirement-ids.js';
+import { softSkip } from '../lib/soft-skip.js';
 
 const DISALLOWED_TOOL = 'conformance-disallowed-tool';
 
@@ -32,12 +33,12 @@ describe('agent-live-allowlist-enforced (RFC 0077 §F-1)', () => {
     const cap = await readLiveRuntimeCap();
     if (!behaviorGate('openwop-live-allowlist-enforced', cap?.supported === true)) return;
 
-    if (!(await isEventLogSeamAvailable())) return; // soft-skip
+    if (!(await isEventLogSeamAvailable())) return softSkip('blocked', 'precondition not met — `!(await isEventLogSeamAvailable())` returned early (soft-skip) (seam, prior step, or fixture unavailable)'); // soft-skip
     const res = await invokeLive({ source: 'run-api', attemptTool: DISALLOWED_TOOL });
-    if (res === null || !res.runId) return; // seam/hook absent — soft-skip
+    if (res === null || !res.runId) return softSkip('blocked', 'precondition not met — `res === null || !res.runId` returned early (seam/hook absent — soft-skip) (seam, prior step, or fixture unavailable)'); // seam/hook absent — soft-skip
 
     const q = await queryTestEvents(res.runId, { type: 'agent.toolCalled' });
-    if (!q.ok) return;
+    if (!q.ok) return softSkip('blocked', 'precondition not met — `!q.ok` returned early (seam, prior step, or fixture unavailable)');
 
     const calledDisallowed = q.events.some((e) => {
       const tool = e.payload.tool ?? e.payload.toolId ?? e.payload.name;
@@ -45,7 +46,7 @@ describe('agent-live-allowlist-enforced (RFC 0077 §F-1)', () => {
     });
     expect(
       calledDisallowed === false,
-      driver.describe('RFC 0077 §F-1 / RFC 0002 §A14', 'a live invocation MUST NOT call a tool outside the agent toolAllowlist'),
+      req('openwop.it.agent-live-allowlist-enforced.does-not-call-a-tool-outside-the-agent-toolallowlist', 'RFC 0077 §F-1 / RFC 0002 §A14', 'a live invocation MUST NOT call a tool outside the agent toolAllowlist'),
     ).toBe(true);
 
     await resetTestSeam();

package/src/scenarios/agent-live-invocation-bracket.test.ts

@@ -21,10 +21,11 @@
  */
 
 import { describe, it, expect } from 'vitest';
-import { driver } from '../lib/driver.js';
 import { behaviorGate } from '../lib/behavior-gate.js';
 import { readLiveRuntimeCap, invokeLive } from '../lib/liveRuntime.js';
 import { queryTestEvents, isEventLogSeamAvailable, resetTestSeam } from '../lib/event-log-query.js';
+import { req } from '../lib/requirement-ids.js';
+import { softSkip } from '../lib/soft-skip.js';
 
 const SOURCES = ['workflow-node', 'run-api', 'chat-mention'];
 const OUTCOMES = ['completed', 'handed-off', 'escalated', 'refused', 'failed'];
@@ -35,21 +36,21 @@ describe('agent-live-invocation-bracket (RFC 0077 §E)', () => {
     const cap = await readLiveRuntimeCap();
     if (!behaviorGate('openwop-live-invocation-bracket', cap?.supported === true)) return;
 
-    if (!(await isEventLogSeamAvailable())) return; // event-log seam absent — soft-skip
+    if (!(await isEventLogSeamAvailable())) return softSkip('blocked', 'precondition not met — `!(await isEventLogSeamAvailable())` returned early (event-log seam absent — soft-skip) (seam, prior step, or fixture unavailable)'); // event-log seam absent — soft-skip
     const res = await invokeLive({ source: 'run-api' });
-    if (res === null || !res.runId) return; // live-invoke seam absent — soft-skip
+    if (res === null || !res.runId) return softSkip('blocked', 'precondition not met — `res === null || !res.runId` returned early (live-invoke seam absent — soft-skip) (seam, prior step, or fixture unavailable)'); // live-invoke seam absent — soft-skip
 
     const q = await queryTestEvents(res.runId);
-    if (!q.ok) return;
+    if (!q.ok) return softSkip('blocked', 'precondition not met — `!q.ok` returned early (seam, prior step, or fixture unavailable)');
     const events = q.events.slice().sort((a, b) => a.sequence - b.sequence);
 
     const started = events.filter((e) => e.type === 'agent.invocation.started');
     const completed = events.filter((e) => e.type === 'agent.invocation.completed');
     expect(
       started.length >= 1 && completed.length >= 1,
-      driver.describe('multi-agent-execution.md §"Live manifest dispatch"', 'a live invocation MUST emit agent.invocation.started + agent.invocation.completed'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'multi-agent-execution.md §"Live manifest dispatch"', 'a live invocation MUST emit agent.invocation.started + agent.invocation.completed'),
     ).toBe(true);
-    if (started.length === 0 || completed.length === 0) return;
+    if (started.length === 0 || completed.length === 0) return softSkip('blocked', 'precondition not met — `started.length === 0 || completed.length === 0` returned early (seam, prior step, or fixture unavailable)');
 
     const start = started[0]!;
     const end = completed[completed.length - 1]!;
@@ -58,11 +59,11 @@ describe('agent-live-invocation-bracket (RFC 0077 §E)', () => {
     const agentScoped = events.filter((e) => AGENT_SCOPED(e.type));
     expect(
       agentScoped[0]?.type === 'agent.invocation.started',
-      driver.describe('RFC 0077 §E', 'agent.invocation.started MUST be the first agent-scoped event of the invocation'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'RFC 0077 §E', 'agent.invocation.started MUST be the first agent-scoped event of the invocation'),
     ).toBe(true);
     expect(
       agentScoped[agentScoped.length - 1]?.type === 'agent.invocation.completed',
-      driver.describe('RFC 0077 §E', 'agent.invocation.completed MUST be the last agent-scoped event of the invocation'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'RFC 0077 §E', 'agent.invocation.completed MUST be the last agent-scoped event of the invocation'),
     ).toBe(true);
 
     // Matching invocationId across the bracket.
@@ -70,17 +71,17 @@ describe('agent-live-invocation-bracket (RFC 0077 §E)', () => {
     const endId = end.payload.invocationId;
     expect(
       typeof startId === 'string' && startId === endId,
-      driver.describe('run-event-payloads.schema.json#agentInvocation*', 'the bracket MUST share one invocationId'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'run-event-payloads.schema.json#agentInvocation*', 'the bracket MUST share one invocationId'),
     ).toBe(true);
 
     // Enum discipline.
     expect(
       typeof start.payload.source === 'string' && SOURCES.includes(start.payload.source as string),
-      driver.describe('run-event-payloads.schema.json#agentInvocationStarted', 'source MUST be workflow-node|run-api|chat-mention'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'run-event-payloads.schema.json#agentInvocationStarted', 'source MUST be workflow-node|run-api|chat-mention'),
     ).toBe(true);
     expect(
       typeof end.payload.outcome === 'string' && OUTCOMES.includes(end.payload.outcome as string),
-      driver.describe('run-event-payloads.schema.json#agentInvocationCompleted', 'outcome MUST be in the closed enum'),
+      req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'run-event-payloads.schema.json#agentInvocationCompleted', 'outcome MUST be in the closed enum'),
     ).toBe(true);
 
     // Content-free: identifiers + metadata only, never prompt/result body.
@@ -88,7 +89,7 @@ describe('agent-live-invocation-bracket (RFC 0077 §E)', () => {
       for (const forbidden of ['prompt', 'result', 'body', 'input', 'output', 'apiKey', 'secret', 'credentials', 'token']) {
         expect(
           !(forbidden in evt.payload),
-          driver.describe('RFC 0077', `agent.invocation.* MUST be content-free (no ${forbidden})`),
+          req('openwop.it.agent-live-invocation-bracket.brackets-a-live-invocation-with-started-first-completed-last-matching-invocation', 'RFC 0077', `agent.invocation.* MUST be content-free (no ${forbidden})`),
         ).toBe(true);
       }
     }

package/src/scenarios/agent-live-runtime-shape.test.ts

@@ -28,9 +28,7 @@ import { join } from 'node:path';
 import Ajv2020 from 'ajv/dist/2020.js';
 import addFormats from 'ajv-formats';
 import { SCHEMAS_DIR } from '../lib/paths.js';
-
-/** Server-free assertion-message helper (mirrors driver.describe's "spec — requirement" shape without requiring OPENWOP_BASE_URL). */
-const why = (specRef: string, requirement: string): string => `${specRef} — ${requirement}`;
+import { req } from '../lib/requirement-ids.js';
 
 function loadSchema(name: string): Record<string, unknown> {
   return JSON.parse(readFileSync(join(SCHEMAS_DIR, name), 'utf8')) as Record<string, unknown>;
@@ -43,12 +41,12 @@ describe('agent-live-runtime-shape: capability advertisement (RFC 0077, server-f
     const live = agents?.properties?.liveRuntime;
     expect(
       live,
-      why('capabilities.md §agents', 'agents.liveRuntime MUST be declared'),
+      req('openwop.it.agent-live-runtime-shape.the-capabilities-schema-declares-agents-liveruntime-with-its-sub-flags', 'capabilities.md §agents', 'agents.liveRuntime MUST be declared'),
     ).toBeDefined();
     for (const flag of ['supported', 'structuredOutput', 'confidenceEscalation', 'sources']) {
       expect(
         live?.properties?.[flag],
-        why('multi-agent-execution.md §Live manifest dispatch', `agents.liveRuntime.${flag} MUST be declared`),
+        req('openwop.it.agent-live-runtime-shape.the-capabilities-schema-declares-agents-liveruntime-with-its-sub-flags', 'multi-agent-execution.md §Live manifest dispatch', `agents.liveRuntime.${flag} MUST be declared`),
       ).toBeDefined();
     }
   });
@@ -64,35 +62,35 @@ describe('agent-live-runtime-shape: invocation event payloads (RFC 0077, server-
   const completed = ajv.getSchema('payloads#/$defs/agentInvocationCompleted');
 
   it('agent.invocation.started validates a content-free start record and requires source', () => {
-    expect(started, 'the agentInvocationStarted $def MUST exist').toBeTruthy();
+    expect(started, req('openwop.it.agent-live-runtime-shape.agent-invocation-started-validates-a-content-free-start-record-and-requires-sour', 'RFC 0077', 'the agentInvocationStarted $def MUST exist')).toBeTruthy();
     expect(
       started!({ invocationId: 'inv-1', agentId: 'vendor.acme.review.code-reviewer', source: 'run-api', modelClass: 'coding', toolSurfaceCount: 3, memoryBound: false }),
-      why('RFC 0077 §C', 'a conforming agent.invocation.started payload MUST validate'),
+      req('openwop.it.agent-live-runtime-shape.agent-invocation-started-validates-a-content-free-start-record-and-requires-sour', 'RFC 0077 §C', 'a conforming agent.invocation.started payload MUST validate'),
     ).toBe(true);
     // Negative: missing source — every invocation must record its entry point.
     expect(
       started!({ invocationId: 'inv-1', agentId: 'vendor.acme.review.code-reviewer' }),
-      why('RFC 0077 §C', 'agent.invocation.started without source MUST be rejected'),
+      req('openwop.it.agent-live-runtime-shape.agent-invocation-started-validates-a-content-free-start-record-and-requires-sour', 'RFC 0077 §C', 'agent.invocation.started without source MUST be rejected'),
     ).toBe(false);
   });
 
   it('agent.invocation.completed validates a content-free outcome record and pins the outcome enum', () => {
-    expect(completed, 'the agentInvocationCompleted $def MUST exist').toBeTruthy();
+    expect(completed, req('openwop.it.agent-live-runtime-shape.agent-invocation-completed-validates-a-content-free-outcome-record-and-pins-the', 'RFC 0077', 'the agentInvocationCompleted $def MUST exist')).toBeTruthy();
     expect(
       completed!({ invocationId: 'inv-1', agentId: 'vendor.acme.review.code-reviewer', outcome: 'completed', schemaValidated: true, confidence: 0.91 }),
-      why('RFC 0077 §C', 'a conforming agent.invocation.completed payload MUST validate'),
+      req('openwop.it.agent-live-runtime-shape.agent-invocation-completed-validates-a-content-free-outcome-record-and-pins-the', 'RFC 0077 §C', 'a conforming agent.invocation.completed payload MUST validate'),
     ).toBe(true);
     // Negative: out-of-enum outcome — the canonical value is `completed`, not `done`.
     expect(
       completed!({ invocationId: 'inv-1', agentId: 'a', outcome: 'done' }),
-      why('RFC 0077 §C', 'agent.invocation.completed with an out-of-enum outcome MUST be rejected'),
+      req('openwop.it.agent-live-runtime-shape.agent-invocation-completed-validates-a-content-free-outcome-record-and-pins-the', 'RFC 0077 §C', 'agent.invocation.completed with an out-of-enum outcome MUST be rejected'),
     ).toBe(false);
   });
 
   it('both invocation event names appear in the RunEventType enum', () => {
     const runEvent = loadSchema('run-event.schema.json');
     const enumVals = (runEvent.$defs as Record<string, { enum?: string[] }>).RunEventType?.enum ?? [];
-    expect(enumVals).toContain('agent.invocation.started');
+    expect(enumVals, req('openwop.it.agent-live-runtime-shape.both-invocation-event-names-appear-in-the-runeventtype-enum', 'RFC 0077', 'both invocation event names appear in the RunEventType enum')).toContain('agent.invocation.started');
     expect(enumVals).toContain('agent.invocation.completed');
   });
 });

package/src/scenarios/agent-live-structured-output.test.ts

@@ -20,10 +20,11 @@
  */
 
 import { describe, it, expect } from 'vitest';
-import { driver } from '../lib/driver.js';
 import { behaviorGate } from '../lib/behavior-gate.js';
 import { readLiveRuntimeCap, invokeLive } from '../lib/liveRuntime.js';
 import { queryTestEvents, isEventLogSeamAvailable, resetTestSeam } from '../lib/event-log-query.js';
+import { req } from '../lib/requirement-ids.js';
+import { softSkip } from '../lib/soft-skip.js';
 
 describe('agent-live-structured-output (RFC 0077 §B step 6)', () => {
   it('fails the invocation on a result that violates handoff.returnSchemaRef', async () => {
@@ -32,25 +33,25 @@ describe('agent-live-structured-output (RFC 0077 §B step 6)', () => {
     const advertised = cap?.supported === true && cap?.structuredOutput === true;
     if (!behaviorGate('openwop-live-structured-output', advertised)) return;
 
-    if (!(await isEventLogSeamAvailable())) return; // soft-skip
+    if (!(await isEventLogSeamAvailable())) return softSkip('blocked', 'precondition not met — `!(await isEventLogSeamAvailable())` returned early (soft-skip) (seam, prior step, or fixture unavailable)'); // soft-skip
     const res = await invokeLive({
       source: 'run-api',
       returnSchemaRef: 'conformance-strict-handoff',
       forceInvalidResult: true,
     });
-    if (res === null || !res.runId) return; // seam/hook absent — soft-skip
+    if (res === null || !res.runId) return softSkip('blocked', 'precondition not met — `res === null || !res.runId` returned early (seam/hook absent — soft-skip) (seam, prior step, or fixture unavailable)'); // seam/hook absent — soft-skip
 
     const q = await queryTestEvents(res.runId, { type: 'agent.invocation.completed' });
-    if (!q.ok || !q.events[0]) return;
+    if (!q.ok || !q.events[0]) return softSkip('blocked', 'precondition not met — `!q.ok || !q.events[0]` returned early (seam, prior step, or fixture unavailable)');
     const payload = q.events[q.events.length - 1]!.payload;
 
     expect(
       payload.outcome === 'failed',
-      driver.describe('RFC 0077 §B step 6', 'a result violating handoff.returnSchemaRef MUST fail the invocation (outcome "failed"), not ship as completed'),
+      req('openwop.it.agent-live-structured-output.fails-the-invocation-on-a-result-that-violates-handoff-returnschemaref', 'RFC 0077 §B step 6', 'a result violating handoff.returnSchemaRef MUST fail the invocation (outcome "failed"), not ship as completed'),
     ).toBe(true);
     expect(
       payload.schemaValidated !== true,
-      driver.describe('RFC 0077 §B step 6', 'schemaValidated MUST NOT be true for a schema-violating result'),
+      req('openwop.it.agent-live-structured-output.fails-the-invocation-on-a-result-that-violates-handoff-returnschemaref', 'RFC 0077 §B step 6', 'schemaValidated MUST NOT be true for a schema-violating result'),
     ).toBe(true);
 
     await resetTestSeam();

package/src/scenarios/agent-loop-iteration-monotonic.test.ts

@@ -11,8 +11,8 @@
 
 import { describe, it, expect } from 'vitest';
 import { softSkip } from '../lib/soft-skip.js';
-import { driver } from '../lib/driver.js';
 import { readExecutionModelCap, isVersion5, invokeAgentLoop } from '../lib/agentLoop.js';
+import { req } from '../lib/requirement-ids.js';
 
 describe('agent-loop-iteration-monotonic (RFC 0061 §B)', () => {
   it('iteration increments by exactly 1 per orchestrator turn, 1-based', async () => {
@@ -22,13 +22,13 @@ describe('agent-loop-iteration-monotonic (RFC 0061 §B)', () => {
     const decisions = res.decisions ?? [];
     expect(
       decisions.length >= 1,
-      driver.describe('RFC 0061 §B', 'a multi-turn loop MUST emit one runOrchestrator.decided per turn'),
+      req('openwop.it.agent-loop-iteration-monotonic.iteration-increments-by-exactly-1-per-orchestrator-turn-1-based', 'RFC 0061 §B', 'a multi-turn loop MUST emit one runOrchestrator.decided per turn'),
     ).toBe(true);
     const iterations = decisions.map((d) => d.iteration);
     const expected = decisions.map((_, k) => k + 1);
     expect(
       JSON.stringify(iterations),
-      driver.describe('RFC 0061 §B', 'iteration MUST be 1-based + monotonic, incrementing by exactly 1 per turn'),
+      req('openwop.it.agent-loop-iteration-monotonic.iteration-increments-by-exactly-1-per-orchestrator-turn-1-based', 'RFC 0061 §B', 'iteration MUST be 1-based + monotonic, incrementing by exactly 1 per turn'),
     ).toBe(JSON.stringify(expected));
   });
 });

package/src/scenarios/agent-loop-stateful-resume.test.ts

@@ -11,8 +11,8 @@
 
 import { describe, it, expect } from 'vitest';
 import { softSkip } from '../lib/soft-skip.js';
-import { driver } from '../lib/driver.js';
 import { readExecutionModelCap, invokeAgentLoop } from '../lib/agentLoop.js';
+import { req } from '../lib/requirement-ids.js';
 
 describe('agent-loop-stateful-resume (RFC 0061 §D)', () => {
   it('a mid-loop suspend resumes at the same iteration, counter intact', async () => {
@@ -23,7 +23,7 @@ describe('agent-loop-stateful-resume (RFC 0061 §D)', () => {
     if (res === null) return softSkip('blocked', 'seam absent — soft-skip');
     expect(
       res.resumedIteration,
-      driver.describe('RFC 0061 §D', 'a stateful resume MUST continue at the suspend iteration — the counter does not reset or skip'),
+      req('openwop.it.agent-loop-stateful-resume.a-mid-loop-suspend-resumes-at-the-same-iteration-counter-intact', 'RFC 0061 §D', 'a stateful resume MUST continue at the suspend iteration — the counter does not reset or skip'),
     ).toBe(2);
   });
 });

package/src/scenarios/agent-loop-version5-shape.test.ts

@@ -13,9 +13,9 @@
 
 import { describe, it, expect } from 'vitest';
 import { softSkip } from '../lib/soft-skip.js';
-import { driver } from '../lib/driver.js';
 import { executionModelVersionMax } from '../lib/execution-model-version.js';
 import { readExecutionModelCap } from '../lib/agentLoop.js';
+import { req } from '../lib/requirement-ids.js';
 
 describe('agent-loop-version5-shape: advertisement (RFC 0061 §A)', () => {
   it('executionModel.statefulResume/transcriptWindow are well-formed when present', async () => {
@@ -24,13 +24,13 @@ describe('agent-loop-version5-shape: advertisement (RFC 0061 §A)', () => {
     if (em.statefulResume !== undefined) {
       expect(
         typeof em.statefulResume,
-        driver.describe('capabilities.schema.json §multiAgent.executionModel', 'statefulResume MUST be a boolean when present'),
+        req('openwop.it.agent-loop-version5-shape.executionmodel-statefulresume-transcriptwindow-are-well-formed-when-present', 'capabilities.schema.json §multiAgent.executionModel', 'statefulResume MUST be a boolean when present'),
       ).toBe('boolean');
     }
     if (em.transcriptWindow !== undefined) {
       expect(
         typeof em.transcriptWindow === 'number' && (em.transcriptWindow as number) >= 1,
-        driver.describe('capabilities.schema.json §multiAgent.executionModel', 'transcriptWindow MUST be a positive integer when present'),
+        req('openwop.it.agent-loop-version5-shape.executionmodel-statefulresume-transcriptwindow-are-well-formed-when-present', 'capabilities.schema.json §multiAgent.executionModel', 'transcriptWindow MUST be a positive integer when present'),
       ).toBe(true);
     }
     if (typeof em.version === 'number') {
@@ -40,7 +40,7 @@ describe('agent-loop-version5-shape: advertisement (RFC 0061 §A)', () => {
       const max = executionModelVersionMax();
       expect(
         (em.version as number) >= 1 && (em.version as number) <= max,
-        driver.describe('capabilities.schema.json §multiAgent.executionModel', `version MUST be within the 1–${max} ladder (schema maximum)`),
+        req('openwop.it.agent-loop-version5-shape.executionmodel-statefulresume-transcriptwindow-are-well-formed-when-present', 'capabilities.schema.json §multiAgent.executionModel', `version MUST be within the 1–${max} ladder (schema maximum)`),
       ).toBe(true);
     }
   });

package/src/scenarios/agent-loop-workspace-snapshot.test.ts

@@ -12,8 +12,8 @@
 
 import { describe, it, expect } from 'vitest';
 import { softSkip } from '../lib/soft-skip.js';
-import { driver } from '../lib/driver.js';
 import { readExecutionModelCap, isVersion5, hasWorkspace, invokeAgentLoop } from '../lib/agentLoop.js';
+import { req } from '../lib/requirement-ids.js';
 
 describe('agent-loop-workspace-snapshot (RFC 0061 §C)', () => {
   it('a turn-i workspace write is invisible to turn i, visible to turn i+1', async () => {
@@ -24,11 +24,11 @@ describe('agent-loop-workspace-snapshot (RFC 0061 §C)', () => {
     const vis = res.workspaceVisible ?? {};
     expect(
       vis.atWriteTurn,
-      driver.describe('RFC 0061 §C', 'a workspace write during turn i MUST be invisible to turn i\'s snapshot'),
+      req('openwop.it.agent-loop-workspace-snapshot.a-turn-i-workspace-write-is-invisible-to-turn-i-visible-to-turn-i-1', 'RFC 0061 §C', 'a workspace write during turn i MUST be invisible to turn i\'s snapshot'),
     ).toBe(false);
     expect(
       vis.atNextTurn,
-      driver.describe('RFC 0061 §C', 'a workspace write during turn i MUST be visible to turn i+1'),
+      req('openwop.it.agent-loop-workspace-snapshot.a-turn-i-workspace-write-is-invisible-to-turn-i-visible-to-turn-i-1', 'RFC 0061 §C', 'a workspace write during turn i MUST be visible to turn i+1'),
     ).toBe(true);
   });
 });

package/src/scenarios/agent-manifest-role-profile.test.ts

@@ -34,9 +34,9 @@ import { join } from 'node:path';
 import Ajv2020 from 'ajv/dist/2020.js';
 import addFormats from 'ajv-formats';
 import { SCHEMAS_DIR } from '../lib/paths.js';
+import { req } from '../lib/requirement-ids.js';
 
 const BASE = 'https://openwop.dev/spec/v1/';
-const why = (specRef: string, requirement: string): string => `${specRef} — ${requirement}`;
 function loadSchema(name: string): Record<string, unknown> {
   return JSON.parse(readFileSync(join(SCHEMAS_DIR, name), 'utf8')) as Record<string, unknown>;
 }
@@ -61,56 +61,56 @@ describe('agent-manifest-role-profile: AgentManifest.role + Skill profile (RFC 0
   it('NO role + any memoryShape (conversation+longTerm) validates — explicit, never inferred', () => {
     expect(
       manifest({ ...base, memoryShape: { scratchpad: true, conversation: true, longTerm: true } }),
-      why('RFC 0131 §A', 'absent role ⇒ unconstrained; nothing reclassifies (today’s meaning, unchanged)'),
+      req('openwop.it.agent-manifest-role-profile.no-role-any-memoryshape-conversation-longterm-validates-explicit-never-inferred', 'RFC 0131 §A', 'absent role ⇒ unconstrained; nothing reclassifies (today’s meaning, unchanged)'),
     ).toBe(true);
   });
 
   it('NO role + handoff + rich memory validates — handoff presence does NOT imply skill', () => {
     expect(
       manifest({ ...base, handoff, memoryShape: { conversation: true, longTerm: true } }),
-      why('RFC 0131 §A / Motivation 2', 'handoff is an interop contract, orthogonal to role — no inference'),
+      req('openwop.it.agent-manifest-role-profile.no-role-handoff-rich-memory-validates-handoff-presence-does-not-imply-skill', 'RFC 0131 §A / Motivation 2', 'handoff is an interop contract, orthogonal to role — no inference'),
     ).toBe(true);
   });
 
   it('role:"skill" WITH handoff + scratchpad-only memory validates', () => {
     expect(
       manifest({ ...base, role: 'skill', handoff, memoryShape: { scratchpad: true, conversation: false, longTerm: false } }),
-      why('RFC 0131 §B', 'a well-formed skill (handoff + scratchpad-only) MUST validate'),
+      req('openwop.it.agent-manifest-role-profile.role-skill-with-handoff-scratchpad-only-memory-validates', 'RFC 0131 §B', 'a well-formed skill (handoff + scratchpad-only) MUST validate'),
     ).toBe(true);
   });
 
   it('role:"skill" with memoryShape.longTerm:true FAILS validation (malformed, not degraded)', () => {
     expect(
       manifest({ ...base, role: 'skill', handoff, memoryShape: { scratchpad: true, longTerm: true } }),
-      why('RFC 0131 §B', 'a skill declaring longTerm memory is malformed — reject at publish/install'),
+      req('openwop.it.agent-manifest-role-profile.role-skill-with-memoryshape-longterm-true-fails-validation-malformed-not-degrade', 'RFC 0131 §B', 'a skill declaring longTerm memory is malformed — reject at publish/install'),
     ).toBe(false);
   });
 
   it('role:"skill" with memoryShape.conversation:true FAILS validation', () => {
     expect(
       manifest({ ...base, role: 'skill', handoff, memoryShape: { conversation: true } }),
-      why('RFC 0131 §B', 'a skill declaring conversation memory is malformed — reject'),
+      req('openwop.it.agent-manifest-role-profile.role-skill-with-memoryshape-conversation-true-fails-validation', 'RFC 0131 §B', 'a skill declaring conversation memory is malformed — reject'),
     ).toBe(false);
   });
 
   it('role:"skill" MISSING handoff FAILS validation', () => {
     expect(
       manifest({ ...base, role: 'skill', memoryShape: { scratchpad: true } }),
-      why('RFC 0131 §B', 'a skill MUST declare handoff (task→return capability)'),
+      req('openwop.it.agent-manifest-role-profile.role-skill-missing-handoff-fails-validation', 'RFC 0131 §B', 'a skill MUST declare handoff (task→return capability)'),
     ).toBe(false);
   });
 
   it('role:"assistant" with conversation+longTerm (and handoff) validates — no profile binds it', () => {
     expect(
       manifest({ ...base, role: 'assistant', handoff, memoryShape: { scratchpad: true, conversation: true, longTerm: true } }),
-      why('RFC 0131 §A', 'an assistant’s conversation + long-term memory are legitimate; it MAY ship handoff'),
+      req('openwop.it.agent-manifest-role-profile.role-assistant-with-conversation-longterm-and-handoff-validates-no-profile-binds', 'RFC 0131 §A', 'an assistant’s conversation + long-term memory are legitimate; it MAY ship handoff'),
     ).toBe(true);
   });
 
   it('role outside the ["skill","assistant"] enum FAILS', () => {
     expect(
       manifest({ ...base, role: 'agent' }),
-      why('RFC 0131 §A', '"agent" is not an enum value (it is the overloaded word the RFC removes)'),
+      req('openwop.it.agent-manifest-role-profile.role-outside-the-skill-assistant-enum-fails', 'RFC 0131 §A', '"agent" is not an enum value (it is the overloaded word the RFC removes)'),
     ).toBe(false);
   });
 });