@openwop/openwop-conformance 1.163.0 → 2.0.0-rc.1

package/src/scenarios/agent-deployment-shape.test.ts

@@ -38,9 +38,7 @@ 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. */
-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>;
@@ -51,11 +49,11 @@ describe('agent-deployment-shape: capability advertisement (RFC 0082, server-fre
     const caps = loadSchema('capabilities.schema.json');
     const agents = (caps.properties as Record<string, { properties?: Record<string, { properties?: Record<string, unknown> }> }>).agents;
     const deployment = agents?.properties?.deployment;
-    expect(deployment, why('capabilities.md §agents', 'agents.deployment MUST be declared')).toBeDefined();
+    expect(deployment, req('openwop.it.agent-deployment-shape.the-capabilities-schema-declares-agents-deployment-with-its-sub-flags', 'capabilities.md §agents', 'agents.deployment MUST be declared')).toBeDefined();
     for (const flag of ['supported', 'channels', 'canary', 'rollback', 'states']) {
       expect(
         deployment?.properties?.[flag],
-        why('agent-deployment.md §F', `agents.deployment.${flag} MUST be declared`),
+        req('openwop.it.agent-deployment-shape.the-capabilities-schema-declares-agents-deployment-with-its-sub-flags', 'agent-deployment.md §F', `agents.deployment.${flag} MUST be declared`),
       ).toBeDefined();
     }
   });
@@ -69,16 +67,16 @@ describe('agent-deployment-shape: deployment record + AgentRef binding (RFC 0082
 
   it('AgentDeployment validates a conforming record and rejects a bad state / out-of-range canary', () => {
     const good = { agentId: 'core.openwop.agents.support-resolver', version: '2.4.0', state: 'active', canaryPercent: 10, channels: ['stable'] };
-    expect(record(good), why('RFC 0082 §C', 'a conforming deployment record MUST validate')).toBe(true);
-    expect(record({ ...good, state: 'live' }), why('RFC 0082 §C', 'an out-of-enum state MUST be rejected')).toBe(false);
-    expect(record({ ...good, canaryPercent: 150 }), why('RFC 0082 §C', 'canaryPercent > 100 MUST be rejected')).toBe(false);
+    expect(record(good), req('openwop.it.agent-deployment-shape.agentdeployment-validates-a-conforming-record-and-rejects-a-bad-state-out-of-ran', 'RFC 0082 §C', 'a conforming deployment record MUST validate')).toBe(true);
+    expect(record({ ...good, state: 'live' }), req('openwop.it.agent-deployment-shape.agentdeployment-validates-a-conforming-record-and-rejects-a-bad-state-out-of-ran', 'RFC 0082 §C', 'an out-of-enum state MUST be rejected')).toBe(false);
+    expect(record({ ...good, canaryPercent: 150 }), req('openwop.it.agent-deployment-shape.agentdeployment-validates-a-conforming-record-and-rejects-a-bad-state-out-of-ran', 'RFC 0082 §C', 'canaryPercent > 100 MUST be rejected')).toBe(false);
   });
 
   it('AgentRef channel XOR version: each alone and neither validate; both is rejected (RFC 0082 §A)', () => {
-    expect(agentRef({ agentId: 'core.x.y.z', version: '1.0.0' }), why('RFC 0082 §A', 'version-only AgentRef MUST validate')).toBe(true);
-    expect(agentRef({ agentId: 'core.x.y.z', channel: 'stable' }), why('RFC 0082 §A', 'channel-only AgentRef MUST validate')).toBe(true);
-    expect(agentRef({ agentId: 'core.x.y.z' }), why('RFC 0082 §A', 'a ref with neither version nor channel MUST validate (host default)')).toBe(true);
-    expect(agentRef({ agentId: 'core.x.y.z', version: '1.0.0', channel: 'stable' }), why('RFC 0082 §A', 'a ref with BOTH version and channel MUST be rejected')).toBe(false);
+    expect(agentRef({ agentId: 'core.x.y.z', version: '1.0.0' }), req('openwop.it.agent-deployment-shape.agentref-channel-xor-version-each-alone-and-neither-validate-both-is-rejected-rf', 'RFC 0082 §A', 'version-only AgentRef MUST validate')).toBe(true);
+    expect(agentRef({ agentId: 'core.x.y.z', channel: 'stable' }), req('openwop.it.agent-deployment-shape.agentref-channel-xor-version-each-alone-and-neither-validate-both-is-rejected-rf', 'RFC 0082 §A', 'channel-only AgentRef MUST validate')).toBe(true);
+    expect(agentRef({ agentId: 'core.x.y.z' }), req('openwop.it.agent-deployment-shape.agentref-channel-xor-version-each-alone-and-neither-validate-both-is-rejected-rf', 'RFC 0082 §A', 'a ref with neither version nor channel MUST validate (host default)')).toBe(true);
+    expect(agentRef({ agentId: 'core.x.y.z', version: '1.0.0', channel: 'stable' }), req('openwop.it.agent-deployment-shape.agentref-channel-xor-version-each-alone-and-neither-validate-both-is-rejected-rf', 'RFC 0082 §A', 'a ref with BOTH version and channel MUST be rejected')).toBe(false);
   });
 });
 
@@ -94,29 +92,29 @@ describe('agent-deployment-shape: deployment.* event payloads (RFC 0082, server-
   const stateChanged = ajv.getSchema('payloads#/$defs/deploymentStateChanged');
 
   it('deployment.promoted validates a content-free promotion record and requires toVersion + toState', () => {
-    expect(promoted, 'the deploymentPromoted $def MUST exist').toBeTruthy();
+    expect(promoted, req('openwop.it.agent-deployment-shape.deployment-promoted-validates-a-content-free-promotion-record-and-requires-tover', 'RFC 0082', 'the deploymentPromoted $def MUST exist')).toBeTruthy();
     expect(
       promoted!({ agentId: 'core.openwop.agents.support-resolver', toVersion: '2.4.0', toState: 'active', channel: 'stable', canaryPercent: 10, evalRunId: 'run_abc' }),
-      why('RFC 0082 §D', 'a conforming deployment.promoted payload MUST validate'),
+      req('openwop.it.agent-deployment-shape.deployment-promoted-validates-a-content-free-promotion-record-and-requires-tover', 'RFC 0082 §D', 'a conforming deployment.promoted payload MUST validate'),
     ).toBe(true);
-    expect(promoted!({ agentId: 'a' }), why('RFC 0082 §D', 'deployment.promoted without toVersion/toState MUST be rejected')).toBe(false);
+    expect(promoted!({ agentId: 'a' }), req('openwop.it.agent-deployment-shape.deployment-promoted-validates-a-content-free-promotion-record-and-requires-tover', 'RFC 0082 §D', 'deployment.promoted without toVersion/toState MUST be rejected')).toBe(false);
   });
 
   it('deployment.rolled-back / canary.adjusted / state.changed validate conforming records', () => {
-    expect(rolledBack!({ agentId: 'a', fromVersion: '2.4.0', toVersion: '2.3.1', rollbackPointer: '2.3.1' }), why('RFC 0082 §D', 'a conforming deployment.rolled-back MUST validate')).toBe(true);
-    expect(canary!({ agentId: 'a', version: '2.4.0', fromPercent: 10, toPercent: 50 }), why('RFC 0082 §D', 'a conforming deployment.canary.adjusted MUST validate')).toBe(true);
-    expect(stateChanged!({ agentId: 'a', version: '2.4.0', fromState: 'active', toState: 'paused' }), why('RFC 0082 §D', 'a conforming deployment.state.changed MUST validate')).toBe(true);
-    expect(stateChanged!({ agentId: 'a', version: '2.4.0', fromState: 'active', toState: 'live' }), why('RFC 0082 §D', 'an out-of-enum toState MUST be rejected')).toBe(false);
+    expect(rolledBack!({ agentId: 'a', fromVersion: '2.4.0', toVersion: '2.3.1', rollbackPointer: '2.3.1' }), req('openwop.it.agent-deployment-shape.deployment-rolled-back-canary-adjusted-state-changed-validate-conforming-records', 'RFC 0082 §D', 'a conforming deployment.rolled-back MUST validate')).toBe(true);
+    expect(canary!({ agentId: 'a', version: '2.4.0', fromPercent: 10, toPercent: 50 }), req('openwop.it.agent-deployment-shape.deployment-rolled-back-canary-adjusted-state-changed-validate-conforming-records', 'RFC 0082 §D', 'a conforming deployment.canary.adjusted MUST validate')).toBe(true);
+    expect(stateChanged!({ agentId: 'a', version: '2.4.0', fromState: 'active', toState: 'paused' }), req('openwop.it.agent-deployment-shape.deployment-rolled-back-canary-adjusted-state-changed-validate-conforming-records', 'RFC 0082 §D', 'a conforming deployment.state.changed MUST validate')).toBe(true);
+    expect(stateChanged!({ agentId: 'a', version: '2.4.0', fromState: 'active', toState: 'live' }), req('openwop.it.agent-deployment-shape.deployment-rolled-back-canary-adjusted-state-changed-validate-conforming-records', 'RFC 0082 §D', 'an out-of-enum toState MUST be rejected')).toBe(false);
   });
 
   it('deployment.* events are content-free — a manifest body and a prompt are rejected (deployment-event-no-content-leak)', () => {
     expect(
       promoted!({ agentId: 'a', toVersion: '2.4.0', toState: 'active', manifestBody: '{...}' }),
-      why('SECURITY invariant deployment-event-no-content-leak', 'a deployment.promoted MUST NOT carry a manifest body'),
+      req('openwop.it.agent-deployment-shape.deployment-events-are-content-free-a-manifest-body-and-a-prompt-are-rejected-dep', 'SECURITY invariant deployment-event-no-content-leak', 'a deployment.promoted MUST NOT carry a manifest body'),
     ).toBe(false);
     expect(
       stateChanged!({ agentId: 'a', version: '2.4.0', fromState: 'active', toState: 'paused', prompt: 'system: …' }),
-      why('SECURITY invariant deployment-event-no-content-leak', 'a deployment.state.changed MUST NOT carry prompt content'),
+      req('openwop.it.agent-deployment-shape.deployment-events-are-content-free-a-manifest-body-and-a-prompt-are-rejected-dep', 'SECURITY invariant deployment-event-no-content-leak', 'a deployment.state.changed MUST NOT carry prompt content'),
     ).toBe(false);
   });
 });
@@ -125,15 +123,15 @@ describe('agent-deployment-shape: §B recorded-fact pin + enum (RFC 0082, server
   it('agent.invocation.started carries the additive recorded-fact resolvedAgentVersion / resolvedChannel', () => {
     const payloads = loadSchema('run-event-payloads.schema.json');
     const started = ((payloads.$defs as Record<string, { properties?: Record<string, unknown> }>).agentInvocationStarted)?.properties ?? {};
-    expect(started.resolvedAgentVersion, why('RFC 0082 §B', 'agent.invocation.started.resolvedAgentVersion MUST be declared (the channel pin)')).toBeDefined();
-    expect(started.resolvedChannel, why('RFC 0082 §B', 'agent.invocation.started.resolvedChannel MUST be declared')).toBeDefined();
+    expect(started.resolvedAgentVersion, req('openwop.it.agent-deployment-shape.agent-invocation-started-carries-the-additive-recorded-fact-resolvedagentversion', 'RFC 0082 §B', 'agent.invocation.started.resolvedAgentVersion MUST be declared (the channel pin)')).toBeDefined();
+    expect(started.resolvedChannel, req('openwop.it.agent-deployment-shape.agent-invocation-started-carries-the-additive-recorded-fact-resolvedagentversion', 'RFC 0082 §B', 'agent.invocation.started.resolvedChannel MUST be declared')).toBeDefined();
   });
 
   it('all four deployment 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 ?? [];
     for (const e of ['deployment.promoted', 'deployment.rolled-back', 'deployment.canary.adjusted', 'deployment.state.changed']) {
-      expect(enumVals).toContain(e);
+      expect(enumVals, req('openwop.it.agent-deployment-shape.all-four-deployment-event-names-appear-in-the-runeventtype-enum', 'RFC 0082', 'all four deployment event names appear in the RunEventType enum')).toContain(e);
     }
   });
 });

package/src/scenarios/agent-eval-run.test.ts

@@ -29,7 +29,6 @@ import { readFileSync } from 'node:fs';
 import { join } from 'node:path';
 import Ajv2020 from 'ajv/dist/2020.js';
 import addFormats from 'ajv-formats';
-import { driver } from '../lib/driver.js';
 import { behaviorGate } from '../lib/behavior-gate.js';
 import { SCHEMAS_DIR } from '../lib/paths.js';
 import {
@@ -39,6 +38,8 @@ import {
   EVAL_CONTENT_FORBIDDEN,
 } from '../lib/agentEval.js';
 import { queryTestEvents, requireEvents, isEventLogSeamAvailable, resetTestSeam } from '../lib/event-log-query.js';
+import { req } from '../lib/requirement-ids.js';
+import { softSkip } from '../lib/soft-skip.js';
 
 function loadSchema(name: string): Record<string, unknown> {
   return JSON.parse(readFileSync(join(SCHEMAS_DIR, name), 'utf8')) as Record<string, unknown>;
@@ -48,7 +49,7 @@ function expectContentFree(payload: Record<string, unknown>, where: string): voi
   for (const f of EVAL_CONTENT_FORBIDDEN) {
     expect(
       !(f in payload),
-      driver.describe('RFC 0081 §C (eval-summary-no-content-leak)', `${where} MUST be content-free (no ${f})`),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'RFC 0081 §C (eval-summary-no-content-leak)', `${where} MUST be content-free (no ${f})`),
     ).toBe(true);
   }
 }
@@ -57,10 +58,10 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
   it('emits eval.started → per-task eval.scored → eval.completed and serves a content-free EvalSummary', async () => {
     const cap = await readEvalSuiteCap();
     if (!behaviorGate('openwop-eval-run', 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 run = await driveEvalRun({ modes: ['golden'] });
-    if (run === null) return; // eval-run seam unwired — soft-skip the whole behavioral suite
+    if (run === null) return softSkip('blocked', 'precondition not met — `run === null` returned early (eval-run seam unwired — soft-skip the whole behavioral suite) (seam, prior step, or fixture unavailable)'); // eval-run seam unwired — soft-skip the whole behavioral suite
 
     // From here the host has ADVERTISED agents.evalSuite AND wired the eval-run
     // seam — missing evidence is a FAILURE, not a soft-skip. A host claiming the
@@ -68,7 +69,7 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     // normative EvalSummary, or it is advertising a capability it doesn't deliver.
     expect(
       typeof run.runId === 'string' && run.runId.length > 0,
-      driver.describe('agent-evaluation.md §B', 'a wired eval-run seam MUST return the projected runId'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §B', 'a wired eval-run seam MUST return the projected runId'),
     ).toBe(true);
     const runId = run.runId as string;
 
@@ -87,15 +88,15 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     // ≥1 eval.scored — a wired eval run MUST emit the full sequence.
     expect(
       startedEvents.length === 1,
-      driver.describe('agent-evaluation.md §C', 'an eval run MUST emit exactly one eval.started'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'an eval run MUST emit exactly one eval.started'),
     ).toBe(true);
     expect(
       scoredEvents.length >= 1,
-      driver.describe('agent-evaluation.md §C', 'an eval run MUST emit at least one eval.scored'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'an eval run MUST emit at least one eval.scored'),
     ).toBe(true);
     expect(
       completedEvents.length === 1,
-      driver.describe('agent-evaluation.md §C', 'an eval run MUST emit exactly one eval.completed'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'an eval run MUST emit exactly one eval.completed'),
     ).toBe(true);
     const started = startedEvents[0]!;
     const completed = completedEvents[0]!;
@@ -104,22 +105,22 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     for (const s of scoredEvents) {
       expect(
         started.sequence < s.sequence,
-        driver.describe('agent-evaluation.md §C', 'eval.started MUST precede every eval.scored'),
+        req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'eval.started MUST precede every eval.scored'),
       ).toBe(true);
       expect(
         s.sequence < completed.sequence,
-        driver.describe('agent-evaluation.md §C', 'every eval.scored MUST precede eval.completed'),
+        req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'every eval.scored MUST precede eval.completed'),
       ).toBe(true);
     }
 
     // One eval.scored per task (count == eval.completed.taskCount).
     expect(
       typeof completed.payload.taskCount === 'number',
-      driver.describe('run-event-payloads.schema.json#/$defs/evalCompleted', 'eval.completed MUST carry a numeric taskCount'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'run-event-payloads.schema.json#/$defs/evalCompleted', 'eval.completed MUST carry a numeric taskCount'),
     ).toBe(true);
     expect(
       scoredEvents.length === completed.payload.taskCount,
-      driver.describe('agent-evaluation.md §C', 'one eval.scored per task (count == eval.completed.taskCount)'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'one eval.scored per task (count == eval.completed.taskCount)'),
     ).toBe(true);
 
     // Content-free (§C / eval-summary-no-content-leak) + score ∈ 0..1, passed boolean.
@@ -129,11 +130,11 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
       expectContentFree(s.payload, 'eval.scored');
       expect(
         typeof s.payload.score === 'number' && (s.payload.score as number) >= 0 && (s.payload.score as number) <= 1,
-        driver.describe('run-event-payloads.schema.json#/$defs/evalScored', 'eval.scored.score MUST be in 0..1'),
+        req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'run-event-payloads.schema.json#/$defs/evalScored', 'eval.scored.score MUST be in 0..1'),
       ).toBe(true);
       expect(
         typeof s.payload.passed === 'boolean',
-        driver.describe('run-event-payloads.schema.json#/$defs/evalScored', 'eval.scored.passed MUST be a boolean'),
+        req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'run-event-payloads.schema.json#/$defs/evalScored', 'eval.scored.passed MUST be a boolean'),
       ).toBe(true);
     }
 
@@ -141,7 +142,7 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     const { status, summary } = await getEvalSummary(runId);
     expect(
       status === 200 && summary !== undefined,
-      driver.describe('agent-evaluation.md §C', `GET /v1/runs/{runId}/eval-summary MUST serve a 200 EvalSummary for a completed eval run (got ${status})`),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', `GET /v1/runs/{runId}/eval-summary MUST serve a 200 EvalSummary for a completed eval run (got ${status})`),
     ).toBe(true);
     const sum = summary as Record<string, unknown>;
     const ajv = new Ajv2020({ strict: false, allErrors: true });
@@ -149,7 +150,7 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     const validate = ajv.compile(loadSchema('eval-summary.schema.json'));
     expect(
       validate(sum),
-      driver.describe('eval-summary.schema.json', `EvalSummary MUST be schema-valid (${ajv.errorsText(validate.errors)})`),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'eval-summary.schema.json', `EvalSummary MUST be schema-valid (${ajv.errorsText(validate.errors)})`),
     ).toBe(true);
 
     const tasks = (sum.tasks as Array<Record<string, unknown>> | undefined) ?? [];
@@ -157,11 +158,11 @@ describe('agent-eval-run (RFC 0081 §B/§C)', () => {
     const taskCount = sum.taskCount as number | undefined;
     expect(
       typeof passedCount === 'number' && typeof taskCount === 'number',
-      driver.describe('eval-summary.schema.json', 'EvalSummary MUST carry numeric passedCount + taskCount'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'eval-summary.schema.json', 'EvalSummary MUST carry numeric passedCount + taskCount'),
     ).toBe(true);
     expect(
       (passedCount as number) <= (taskCount as number),
-      driver.describe('agent-evaluation.md §C', 'EvalSummary.passedCount MUST NOT exceed taskCount'),
+      req('openwop.it.agent-eval-run.emits-eval-started-per-task-eval-scored-eval-completed-and-serves-a-content-free', 'agent-evaluation.md §C', 'EvalSummary.passedCount MUST NOT exceed taskCount'),
     ).toBe(true);
     for (const t of tasks) {
       expectContentFree(t, 'EvalSummary.tasks[]');

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');
   });
 });