Live · Updated every 15 minutes

Accuracy Ledger

Every study GodlyDeeds generates is checked against scripture, labeled by interpretive method, and audited for scholarly balance. The numbers below are live. When we get something wrong, it appears in the correction log.

Last 30 days · 79 studies generated

Fact-check pass rate

≥ 95%

100.00%

Share of studies whose cited scripture references were verified line-by-line against the canonical text.

Content-filter flag rate

≤ 2%

6.59%

Share of generations flagged by content moderation. Lower is better.

Parse-error rate

≤ 1%

0.00%

Share of studies that failed structural validation and were regenerated automatically.

User-reported corrections (30 days)

Total reported

0

Resolved

100%

Correction rate

0.00%

Per 1,000 studies

0.0

Factual0

0.0/1k studies

Citation0

0.0/1k studies

Doctrinal0

0.0/1k studies

Method0

0.0/1k studies

Balance0

0.0/1k studies

Recent corrections

No corrections reported yet. When users flag an error, it will appear here.

Active study methodology

Every study is generated using a version-controlled research protocol. When we improve the protocol, the version number increments and the change note is recorded here. This is the audit trail — nothing is hidden.

bible_study_system v2

v2: Added SPIRITUAL POSTURE (Holy Spirit as true teacher, AI as tool not idol) + CHARACTER REFERENCE VARIETY (fix "the psalmist" repetition in Psalms)

Activated 4/6/2026

chat_system v1

v1 seed — source: app/api/chat/route.ts SYSTEM_PROMPT

Activated 4/3/2026

godly_deeds_system v1

v1 seed — source: app/prompts/scriptureInterpretationPrompts.ts GODLY_DEEDS_SYSTEM_PROMPT

Activated 4/3/2026

heart_of_story_system v1

v1 seed — source: app/prompts/heartOfStoryPrompt.ts heartOfStoryPrompt.systemPrompt

Activated 4/3/2026

heart_of_story_user v1

v1 seed — source: app/prompts/heartOfStoryPrompt.ts heartOfStoryUserTemplate. Placeholders: {{topic}}, {{mainScripture_reference}}, {{mainScripture_text}}, {{backstory}}

Activated 4/3/2026

scripture_timeline_system v1

v1 seed — source: app/prompts/timelinePrompt.ts scriptureTimelinePrompt.systemPrompt

Activated 4/5/2026

scripture_timeline_user v1

v1 seed — source: app/prompts/timelinePrompt.ts scriptureTimelineUserTemplate. Placeholders: {{topic}}, {{mainScripture_reference}}, {{mainScripture_text}}, {{backstory}}

Activated 4/5/2026

timeline_system v1

v1 seed — source: app/prompts/timelinePrompt.ts timelinePrompt.systemPrompt

Activated 4/3/2026

timeline_user v1

v1 seed — source: app/prompts/timelinePrompt.ts timelineUserTemplate. Placeholders: {{topic}}, {{mainScripture_reference}}, {{mainScripture_text}}, {{backstory}}, {{heartOfTheStory}}

Activated 4/3/2026

How we define accuracy

  • Scripture grounding. Every scripture reference is verified against the canonical verse text before a study is delivered. Failed verifications trigger regeneration, not silent correction.
  • Interpretive method. Every study declares its interpretive approach — grammatical-historical, typological, canonical, redemptive-historical. Typological readings are only permitted when the New Testament explicitly warrants them.
  • Scholarly balance. Contested theological claims must ship with counter-evidence — the scholars who disagree and why. Studies that fail this check are rewritten, not published.
  • Citation density. Every claim carries at least two evidence items. Claims that cannot be sourced are labeled as speculative or removed.
  • Public corrections. When a user reports an error, it appears on this page. Nothing is quietly patched.
AboutPrivacyHomeSnapshot refreshed every 15 minutes.