feat(regulatory): seed-regulatory-actions.mjs — classify + Redis write

[koala73]

Context

Part of the Regulatory RSS pipeline. Builds on #2492 (fetch + parse layer).

This is task 2/4.

What to build

Add tier classification and Redis write to scripts/seed-regulatory-actions.mjs.

Tier classification — keyword match on title

const HIGH_KEYWORDS = [
  'enforcement', 'charges', 'charged', 'fraud', 'failure', 'failed bank',
  'emergency', 'halt', 'suspension', 'suspended', 'cease', 'desist',
  'penalty', 'fine', 'fined', 'settlement', 'indictment', 'manipulation',
  'ban', 'revocation', 'insolvency',
];

const MEDIUM_KEYWORDS = [
  'proposed rule', 'final rule', 'rulemaking', 'guidance', 'warning',
  'notice', 'advisory', 'review', 'examination', 'investigation',
  'stress test', 'capital requirement', 'disclosure requirement',
];
// Everything else → 'low'

Output payload written to regulatory:actions:v1

{
  "actions": [
    {
      "id": "sec-enforcement-xyz-corp-20260329",
      "agency": "SEC",
      "title": "SEC Charges XYZ Corp with Accounting Fraud",
      "link": "https://www.sec.gov/...",
      "publishedAt": "2026-03-29T14:00:00Z",
      "tier": "high",
      "matchedKeywords": ["charges", "fraud"]
    }
  ],
  "fetchedAt": 1711718400000,
  "recordCount": 12,
  "highCount": 2,
  "mediumCount": 4
}

runSeed integration

• Redis key: regulatory:actions:v1
• TTL: 7200 (2h — matches cron cadence with 1× buffer)
• validateFn: data => Array.isArray(data?.actions) (empty array is valid — no regulatory actions today is fine)
• Domain: 'regulatory', name: 'actions'

Acceptance criteria

• HIGH/MEDIUM/LOW classification correct for sample titles
• matchedKeywords array populated for each action
• Payload shape matches spec above
• runSeed() wired with correct TTL and validate function
• Empty actions array does not fail validation

[lspassos1]

Implemented in #2567.

This adds the classification + Redis publish phase for the regulatory RSS pipeline:

• tier / matchedKeywords classification using the issue keyword sets
• payload aggregation with fetchedAt, recordCount, highCount, and mediumCount
• runSeed('regulatory', 'actions', 'regulatory:actions:v1', ...) with TTL 7200
• expanded unit coverage for classification, payload counts, and publish wiring

This branch is currently stacked on #2564, which provides the fetch/parse base.

Follow-up work remains tracked in #2494 and #2495.

[jaspercross1231-del]

"Looks good! The tier classification based on keywords will make a big difference for sorting actions properly. I’ll keep an eye out for the Redis integration in #2492. The validation for empty actions makes sense too!"

…

On Mon, Mar 30, 2026, 6:40 PM Lucas Passos ***@***.***> wrote: *lspassos1* left a comment (koala73/worldmonitor#2493) <#2493 (comment)> Implemented in #2567 <#2567>. This adds the classification + Redis publish phase for the regulatory RSS pipeline: - tier / matchedKeywords classification using the issue keyword sets - payload aggregation with fetchedAt, recordCount, highCount, and mediumCount - runSeed('regulatory', 'actions', 'regulatory:actions:v1', ...) with TTL 7200 - expanded unit coverage for classification, payload counts, and publish wiring This branch is currently stacked on #2564 <#2564>, which provides the fetch/parse base. Follow-up work remains tracked in #2494 <#2494> and #2495 <#2495>. — Reply to this email directly, view it on GitHub <#2493?email_source=notifications&email_token=CAXZCSTMLWFWGZAR2LNK23L4TMAXZA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMJVHA4DEOBQHA3KM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#issuecomment-4158828086>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CAXZCSSC242OJRIGYC2NVVL4TMAXZAVCNFSM6AAAAACXEQ7HGOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DCNJYHAZDQMBYGY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>