feat(scenarios): complete Tier 1-3 scenario catalog#12
Merged
That1Drifter merged 3 commits intomasterfrom Apr 10, 2026
Merged
Conversation
Fill in all five stub scenarios (doc-qa-rag, pipeline-automation, workflow-agent, legacy-migration, incident-response) with full manifests and intros, plus a stakeholder-conflict surprise in support-triage. - doc-qa-rag (T1): Helix Biotherapeutics RAG buildout, hallucination and restricted-doc leak surprises - pipeline-automation (T2): Keelwater Brokerage carrier feeds with schema drift as the core test - workflow-agent (T2): Axleflow Telematics 3-system onboarding with mid-workflow billing failure - legacy-migration (T3): Ironridge Mutual 17-year-old SOAP gateway, stakeholder archaeology - incident-response (T3): Lumenvest fintech page with turn_budget: 12 - support-triage: launch_vs_testing_conflict surprise (Priya vs Marcus) All manifests validate against scenario.schema.json; pnpm typecheck and pnpm test pass across the workspace. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…atterns Surfaced by staging smoke test: the action_pattern surprise trigger in surprises.ts constructed `new RegExp(pattern)` with no flags and no try/catch, so a manifest pattern using Perl-style `(?i)` (which JS regex doesn't support) crashed the turn handler mid-session. - surprises.ts: build regex with `'i'` flag, wrap in try/catch like score.ts does — a malformed pattern now skips the rule instead of throwing through the request. - score.ts: also add `'i'` flag to payload_regex for consistency with payload_contains, which is already documented case-insensitive. - doc-qa-rag + workflow-agent: strip the now-redundant `(?i)` prefix from six patterns so manifests stay JS-regex-portable. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
TODO.md.launch_vs_testing_conflictstakeholder-conflict surprise tosupport-triage(Priya pushes launch for a quarterly OKR, Marcus pushes for more testing), triggered deterministically whentaxonomy_definedreachesattempted.TODO.mdto mark the six items done.Scenarios
doc-qa-ragphantom_policy_citation,restricted_leak,contradictory_wiki_pagespipeline-automationschema_drift_carrier_renameis the centerpieceworkflow-agentbilling_500_mid_workflow,provisioning_rate_limit,audit_log_mandatelegacy-migrationratebureau_complaint,hector_public_pushback,compliance_audit_noticeincident-responseturn_budget: 12, multi-cause RCA under time pressure;vip_public_complaint,downstream_auth_500s,imani_shift_handoffAll manifests follow
support-triageas the reference structure and stay under the <$2/session cost target.Test plan
pnpm validate-scenarios— 7/7 scenarios parse and validate againstscenario.schema.jsonpnpm typecheck— 6/6 workspace packages cleanpnpm test— 34 tests acrosscore,rubric,scenariosall passpnpm --filter @fieldwork/cli dev -- dryrun <path> --trainee perfectbefore the first real play session🤖 Generated with Claude Code