AI Fortune Telling Prompt Guide for Accurate Readings

AI Fortune Telling Prompt Guide for Accurate Readings: Verified Field Ledger

All prompts must stay within the engine's exported fields, verified on 2026-08-21 from lib/bazi-engine/index.ts. The public ledger includes: four-pillar computation; branch, stem and pillar interactions; strength; pattern; useful-god lenses; shensha; luck cycles and annual/month/day/hour ladders; cross-chart interactions; lunar conversion; and timing ratings. It does not export a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor. A prompt can request those exported fields and must label any prose beyond them as interpretation. This page is an editorial guide, not an interactive prompt tool. For chart-specific output, see the relevant page.

AI Fortune Telling Prompt Guide for Accurate Readings: Input and Pillar-Position Check

Before trusting any reading, verify the raw inputs and the four-pillar positions. The public engine computes the year, month, day, and hour pillars from the birth data you supply; if the hour is missing or the calendar conversion is off, every downstream field shifts. Check that the stem and branch for each pillar match a known reference, then confirm the engine’s exported strength, pattern, and useful-god lenses reflect those pillars. Boundary cases include charts near a solar-term change or a late-night hour where the day pillar may flip. For the mechanics behind these positions, see relevant page. Do not request bone-weight, naming, exam, or deterministic event outputs—those are not exported, and any prose beyond the verified fields is interpretation.

Same-Chart Interaction Comparison: Verification Procedure

When two charts trigger the same interaction label—such as a branch clash or a stem combination—compare them field by field to confirm the boundary. First, verify that each chart exports four-pillar computation, branch, stem and pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles and annual/month/day/hour ladders, cross-chart interactions, lunar conversion, and timing ratings. If any field is missing, the comparison cannot be validated. Next, check that no chart includes a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor; presence of these means the output is not from the public engine. Finally, treat any prose beyond these exported fields as interpretation. For a deeper walkthrough, see relevant page.

Strength and Pattern Verification

Before trusting any reading, check whether the prompt explicitly requests strength (日主 strength) and pattern (格局) fields. These are computed by the public engine, but they are not absolute verdicts. Request the raw output: strength as a class (e.g., strong or weak), pattern as a named classification from the engine, and any useful-god lenses tied to that pattern. Then verify each field against the engine’s exported capabilities. If the response adds a bone-weight score, a naming recommendation, an exam prediction, or a deterministic event claim, flag it as outside the verified boundary. A prompt may ask for strength and pattern, but any prose explaining their life meaning is interpretation, not engine output. See relevant page for background on day-master strength.

To verify a luck-cycle record, check each exported field against the engine output. Start with the four-pillar computation, then confirm branch, stem, and pillar interactions. Next, validate strength, pattern, and useful-god lenses. Record shensha flags, followed by luck cycles and annual/month/day/hour ladders. Include cross-chart interactions and lunar conversion. Finally, note timing ratings. Any prose beyond these exported fields must be labeled as interpretation. If a prompt requests a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor, flag it as out of scope—the public engine does not export those. For further detail on luck cycles, see relevant page.

To verify 'other reported labels' field-by-field, first request the four-pillar computation and lunar conversion, then ask for branch, stem, and pillar interactions. Next, request strength, pattern, and useful-god lenses, followed by shensha, luck cycles, and annual/month/day/hour ladders. Ask for cross-chart interactions and timing ratings to complete coverage. For each label returned, confirm it maps to one of these export names. If a label appears outside this set—such as bone-weight, naming, exam result, or deterministic event prediction—treat the prose as interpretation and label it accordingly. Relevant page defines 'ten gods' as a pillar interaction output, not a standalone prediction. The guide itself is editorial; it does not compute charts. Any listed field must trace back to the engine's public exports, never to an interactive tool or prediction service.

AI Fortune Telling Prompt Guide for Accurate Readings: Missing-Capability Inventory

To verify a prompt's boundary, check each requested field against the public engine's exports. The engine returns four-pillar computation, branch/stem/pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles (annual/month/day/hour), cross-chart interactions, lunar conversion, and timing ratings. If a prompt asks for bone-weight, naming advice, exam outcomes, or any deterministic event prediction, those capabilities are absent from the export list. Mark them as "not available" rather than attempting to simulate them. For any prose beyond the exported fields, label it as interpretation. Consult the relevant page for the engine's documented boundaries. This field-by-field check prevents false claims about what a reading can produce.

Field-by-Field Verification: Separating Engine Output from Interpretation

To distinguish verifiable claims from non-verifiable ones, check each requested field against the engine’s public exports. The engine provides four-pillar computation, branch–stem–pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles and annual/month/day/hour ladders, cross-chart interactions, lunar conversion, and timing ratings. A prompt may request any of these fields; the engine’s return is verifiable. Any prose beyond these fields—such as life predictions, advice, or symbolic meanings—must be labeled as interpretation. Conversely, the engine does not export a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor. If a prompt requests those, the response cannot be verified by the engine and must be marked as non-verifiable or out of scope. The page is an editorial guide, not an interactive prompt tool.

AI Fortune Telling Prompt Guide for Accurate Readings: Reader Audit Procedure

To verify a prompt’s boundaries, check each requested field against this public engine’s exports. Confirm that four-pillar computation, branch/stem/pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles (annual, month, day, hour ladders), cross-chart interactions, lunar conversion, and timing ratings are all explicitly present. If a prompt requests a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor, flag it as outside the engine’s scope. Any prose beyond the exported fields must be labeled as interpretation rather than computed output. Repeat this audit for every field in the prompt; if a request combines exported and non-exported items, separate them before use. This procedure ensures the prompt stays within verified capability boundaries and does not claim unsupported predictions.

Field-by-Field Verification
Compare each requested field against the engine's exported outputs. The public engine returns four-pillar computation, branch/stem/pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles (annual/month/day/hour), cross-chart interactions, lunar conversion, and timing ratings. A prompt may ask for any of these and expect a direct engine result. If a prompt requests a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor, the engine cannot output that field; such prose must be explicitly labeled as interpretation. For boundary cases, list the exported fields first, then note any missing field and mark its absence as outside engine capability. This input-to-output comparison prevents treating interpretive prose as computed data.

Pre-publication fact audit: for each prompt field, verify it maps to an exported engine capability—four-pillar computation; branch, stem, and pillar interactions; strength; pattern; useful-god lenses; shensha; luck cycles and annual/month/day/hour ladders; cross-chart interactions; lunar conversion; timing ratings. Mark any missing field as unavailable. Flag boundary cases: if a request asks for bone-weight calculation, naming recommendations, exam-result prediction, or deterministic event prediction, label it out of scope because the public engine does not export these. For prose beyond exported fields, require an explicit interpretation label. Confirm that no prompt implies the page is an interactive tool or prediction service; it is an editorial guide only. Remove any unverifiable claim, rule, mechanism, or number.

To verify a conclusion, check each field against the engine's exported scope. Four-pillar computation, branch/stem/pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles and annual/month/day/hour ladders, cross-chart interactions, lunar conversion, and timing ratings are allowed. If a prompt requests a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor, mark it out of scope. Label any prose beyond exported fields as interpretation. A boundary case: a request for "luck cycle timing" is valid because luck cycles and timing ratings are exported; a request for "exam outcome" is invalid because the exam-result predictor is not exported. Treat every unexported field as a hard limit, not a soft assumption.

AI Fortune Telling Prompt Guide for Accurate Readings: Verified Field Ledger

To avoid overstating a reading, treat the public engine’s exports as a ledger of verified fields. Before accepting any prompt output, check each requested item against this boundary: four-pillar computation, branch–stem–pillar interactions, strength, pattern, useful-god lenses, shensha, luck cycles with annual/month/day/hour ladders, cross-chart interactions, lunar conversion, and timing ratings. If a prompt asks for a bone-weight calculator, naming recommender, exam-result predictor, or deterministic event predictor, flag it as unsupported by the engine. Any prose beyond these exported fields must be labeled as interpretation, not engine output. This field-by-field verification keeps prompt results aligned with the engine’s actual capability boundary, preventing users from mistaking editorial commentary for computed data.