feat(lenses): Leviticus 21-27 lens content, batch 3 of Pentateuch-rest pilot (#820, #1782)

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s

[CraigBuckmaster]

Summary

Batch 3 of the Pentateuch-rest pilot tracked under #1782. Adds 26 hermeneutic-lens entries across Leviticus 21–27, completing the Leviticus phase. Picks up after #1801 (lev1–10) and #1802 (lev11–20).

After this PR, Leviticus is done. Numbers (36 ch) and Deuteronomy (34 ch) remain to close the pilot.

Per-chapter distribution (26 total)

Chapter	Lenses	Why
lev21 (priestly holiness)	4	priesthood code peak — grammatical, christocentric, typological, canonical
lev22 (priestly purity, sacred food)	2	unblemished-sacrifice + exodus-holiness formula
lev23 (sacred calendar / feasts)	6	maximum — every feast carries a distinct NT fulfillment
lev24 (lampstand, bread, blasphemy)	3	three distinct movements — canonical, typological, grammatical
lev25 (Sabbatical + Jubilee)	5	theology peak — Jubilee + goel + sabbath rest
lev26 (blessings and curses)	4	covenant-treaty form + canon-wide curse template + restoration
lev27 (vows and dedications)	2	sparse appendix — valuation grammar + herem/tithe

Avg 3.7 entries/ch — slightly higher than batches 1–2 (3.2 / 3.3) because three of the seven chapters are theology peaks. Pattern still holds: peaks fully spread, procedural chapters pared back.

Theological highlights

• lev21 (4 lenses): unblemished-priest rule (vv 17–23) anticipating Heb 7:26's Christ "holy, blameless, pure"; high priest restrictions (vv 10–15) as type of unbroken sanctuary access fulfilled in Heb 7–9; v 8's holy-because-they-offer formula echoing into 1 Pet 2:9's royal priesthood.
• lev23 (6 lenses, peak): the seven appointed times mapped to NT fulfillments — Passover→1 Cor 5:7 (christocentric); Firstfruits→1 Cor 15:20–23 (typological); Pentecost-day count→Acts 2 (redemptive); Tabernacles→Jn 7:37–39 / Rev 21:3 (canonical). Devotional anchor on living-in-booths trains repentance from self-sufficiency.
• lev24: vv 1–4 lampstand threading from Zech 4 to Rev 1:12–13; vv 5–9 bread of the Presence anticipating Jn 6:35; the Hebrew naqab wordplay in v 11 ("to pierce" → "to pronounce in curse") for the blasphemy case.
• lev25 (5 lenses, peak): Jubilee proclamation (vv 8–13) cited by Christ at Lk 4:18–19 via Isa 61; the goel kinsman-redeemer (vv 25–28) threading from Boaz to Job 19:25 to Christ; v 18's sixth-year provision as devotional trust; the sojourner-redemption case (vv 47–55) for mission posture toward the nations.
• lev26: five-wave sevenfold curse escalation (vv 14–39) as canonical template Deut 28 expands and the prophets cite throughout exile literature; vv 40–45's patriarchal-covenant memory anchoring restoration beyond Sinai.

Pipeline gate results

schema_validator.py 145762 passed, 0 failed, 19 warnings (pre-existing ESV downloadable warns)
lens_quality_scorer.py 26/26 entries at 100/100 (per-chapter run, floor=100)
build_sqlite.py scripture.db green, hermeneutic_lenses chapter rows = 468
(= 442 baseline from #1802 + 26 batch 3; matches expected)
validate_sqlite.py 101 passed, 0 failed, 2 warnings (pre-existing embeddings/prompts)

Per-chapter SQLite distribution post-build (chapter_lens_content):

lev21: 4 lev22: 2 lev23: 6 lev24: 3 lev25: 5 lev26: 4 lev27: 2

Length discipline

All entries authored to a 250-char target with 30-char headroom against the 280 schema ceiling. Actual range: 201–236 chars. Median ~222.

Filler / token guards

• All six banned filler patterns (the lens reveals, this lens shows, through this lens, this passage shows that, this text teaches that, in this chapter we see/learn) verified absent at authoring time.
• Rubric-token trap (carried forward from batches 1–2): every entry uses a token from its own lens's rubric, not a sibling lens's. Specifically verified: typological entries use type / pattern / prefigures / anticipates; canonical entries use canon / echoes / thread / across Scripture; christocentric entries name Christ / Jesus / Messiah explicitly; mission entries use nations / outward.

Watch list for tier-2 audit

These are the entries most worth a human spot-check during accuracy auditing:

1. lev21 typological — "Heb 7–9 reads them as foreshadowing Christ's once-for-all sanctuary entry without departure." The high-priest-no-leaving-the-sanctuary detail (v 12) as type of unbroken access is a defensible move (Vos, Kline) but the Hebrews argument is more often made from the Day-of-Atonement entry. The link should be sharpened in audit if needed.
2. lev23 redemptive — "the calendar Acts 2 enacts." The Pentecost-fifty-day count from Lev 23:15–16 is the calendrical link to Acts 2:1, but the Acts text itself doesn't quote Lev 23. The connection is canonical-historical rather than citation-explicit. Wording is intentional shorthand.
3. lev24 grammatical — "The Hebrew naqab (v 11) extends from 'to pierce' to 'to pronounce a name in curse'." The semantic extension is well-attested (BDB, HALOT both list both senses; Milgrom develops it in the Anchor Bible volume) but some lexicons split into homonym roots. Defensible either way.
4. lev25 canonical — "Christ as the great Goel." The titular use of Goel-for-Christ is theological synthesis rather than a direct NT citation. Anchor Scriptures (Ruth 4, Job 19:25) are solid; the Christological extension is canonical reading rather than verse-cited.

Out of scope

• No new content generation in content/leviticus/{N}.json itself (only the lens chapter files).
• No app code changes.
• No CI workflow changes.
• app/assets/db-manifest.json and app/assets/explore-images.json drift was checked out before staging.

Rollback

git revert <merge-commit> is sufficient. No schema changes, no migration, no R2 mutation in this PR.

Refs

• Epic: Epic: Hermeneutic Lenses Content #820 (Hermeneutic Lenses)
• Tracker: [#820 / 3] Pentateuch rest — Exodus, Leviticus, Numbers, Deuteronomy (137 ch) #1782 (Pentateuch-rest pilot)
• Prior PRs in path: feat(lenses): Exodus 1-10 lens content, batch 1 of Exodus pilot (#820, #1781) #1797 / feat(lenses): Exodus 11-20 lens content, batch 2 of Exodus pilot (#820, #1781) #1798 / feat(lenses): Exodus 21-30 lens content, batch 3 of Exodus pilot (#820, #1781) #1799 / feat(lenses): Exodus 31-40 lens content, batch 4 of Exodus pilot — closes Exodus (#820, #1781) #1800 (Exodus, all merged), feat(lenses): Leviticus 1-10 lens content, batch 1 of Pentateuch-rest pilot (#820, #1782) #1801 (Leviticus 1–10, merged), feat(lenses): Leviticus 11-20 lens content, batch 2 of Pentateuch-rest pilot (#820, #1782) #1802 (Leviticus 11–20, merged).

Leviticus complete with this PR. Numbers up next.

[github-actions]

Content Pipeline Results

✅ All pipeline checks passed

Step	Status	Details
Schema Validation	✅	145762 passed, 0 failed
Build DB	✅	—
DB Integrity	✅	101 passed, 0 failed

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 4001	❌ 0	4001
Suites	✅ 532	❌ 0	532

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 94.1s