Set Qualification Rules

Purpose

Create a reusable ruleset and decision algorithm. Separate rule authoring from record evaluation and scoring.

Required inputs

• Supplied account and contact inclusion criteria
• Explicit exclusions and terminal consent rules
• Required evidence for each criterion
• Optional record for a requested example evaluation

If a required rule or evidence standard is missing, record it as missing rather than inventing one.

Workflow

1. Reject criteria based on protected traits or unnecessary sensitive personal data, even when supplied. Do not encode or infer them.
2. Convert each lawful, relevant inclusion criterion into an independently checkable rule with PASS and FAIL conditions.
3. Convert each valid exclusion into a hard-disqualification rule whose FAIL forces INELIGIBLE.
4. Define missing, stale, ambiguous, or conflicting required evidence as FAIL and record the missing field.
5. Write the decision algorithm: evaluate hard disqualifiers first; return ELIGIBLE only when every inclusion rule passes and none applies; otherwise return INELIGIBLE.
6. Evaluate only an explicit record when requested. Label it Optional example; otherwise produce no aggregate decision. Normally route direct or batch evaluation to match-icp.
7. Define scoring, if requested, as a later process available only after an ELIGIBLE record decision.

Output format

Then state:

• Decision algorithm: [ordered reusable logic]
• Rejected criteria: [unsafe or invalid supplied rules and reason]

Only when an explicit record is supplied and evaluation is requested, add:

• Optional example — Decision: ELIGIBLE | INELIGIBLE
• Failed rules: [IDs and supplied evidence]
• Missing evidence: [required record facts not supplied]

Guardrails

• Use observable conditions, not vague fit, enthusiasm, or model judgment.
• Reject protected-trait targeting and unnecessary sensitive-data collection or inference, including health or family-status details.
• Use only lawful, necessary, relevant business criteria. Require human and legal review for jurisdiction-sensitive rules.
• Never let a score, reply, or apparent intent offset a failed eligibility rule.
• Do not invent alternate titles, equivalencies, thresholds, exceptions, or evidence.
• Keep facts, inferences, and missing information distinct.
• Do not research, score live records, send outreach, or mutate external systems.

Quality check

Confirm rules return PASS or FAIL independently; unsafe criteria are rejected; jurisdiction-sensitive rules require human and legal review; one hard disqualifier or required missing evidence yields INELIGIBLE; and no decision appears without a record.

Example invocation

Use set-qualification-rules on the Northstar Relay fixture. Encode the company, size, operating-function, employer, role, competitor, and opt-out gates as a reusable ruleset without evaluating a record.