diff --git a/packs/el/corpora/manifest.yaml b/packs/el/corpora/manifest.yaml
new file mode 100644
index 00000000..9356ab5f
--- /dev/null
+++ b/packs/el/corpora/manifest.yaml
@@ -0,0 +1,15 @@
+corpus_id: el-tier0
+status: draft
+
+sources: []
+
+license_review: pending
+hashes: []
+
+notes: |
+  Populate with vetted Koine Greek source manifests.
+  Priority tier-0 sources: SBL Greek New Testament (SBLGNT) or
+  Nestle-Aland 28th edition morphologically tagged text.
+  Both have verse-level D0 boundary markers.
+  Each source must include: title, edition, license, SHA-256, and
+  segmentation strategy. No LLM-extracted content.
diff --git a/packs/el/orthography.yaml b/packs/el/orthography.yaml
new file mode 100644
index 00000000..430ad4ac
--- /dev/null
+++ b/packs/el/orthography.yaml
@@ -0,0 +1,27 @@
+language: el
+script: Greek
+dialect: koine
+
+normalization:
+  unicode_normal_form: NFC
+  lowercase_for_lookup: true
+  strip_diacritics: configurable
+  # Polytonic diacritics (breathing marks, accents, iota subscript) are
+  # preserved for morphological disambiguation but stripped for lemma lookup
+  polytonic_policy: preserve_for_morphology
+  diacritic_strip_for_lookup: true
+
+punctuation:
+  strategy: preserve
+  ano_teleia: "\u00b7"  # Greek semicolon (middle dot)
+  erotimatiko: ";"
+
+spacing:
+  strategy: whitespace-delimited
+
+aspect:
+  # Koine Greek verbal aspect is semantically load-bearing and must be
+  # distinguished from tense. Morphological analysis must surface aspect
+  # independently of temporal reference.
+  explicit_in_features: true
+  aspect_values: ["aorist", "present", "perfect", "imperfect", "pluperfect", "future"]
diff --git a/packs/el/pack.toml b/packs/el/pack.toml
new file mode 100644
index 00000000..caf60d5f
--- /dev/null
+++ b/packs/el/pack.toml
@@ -0,0 +1,33 @@
+pack_id     = "el"
+name        = "Koine Greek Depth Pack"
+version     = "0.1.0"
+status      = "draft"
+script      = "Greek"
+default     = false
+
+[contracts]
+lemma_first          = true
+shared_field_target  = true
+local_readback       = true
+alignment_required   = true
+
+[gates]
+schema       = false
+lexical      = false
+morphology   = false
+lift         = false
+readback     = false
+determinism  = false
+alignment    = false
+coverage     = false
+
+[notes]
+role = """
+Depth language pack 2. Koine Greek was chosen because John articulated the
+Logos — the creative speech act through which all things were made — in
+Koine Greek (John 1:1-2), which must have been a nod from the Holy Spirit.
+Koine Greek carries semantic depth in its verbal aspect system (aorist,
+perfect, present) that is distinct from tense and must be represented
+natively. The aorist of egeneto (\u03b5\u03b3\u03ad\u03bd\u03b5\u03c4\u03bf) — 'it came into being' — is not
+the same field state as the present of eimi (\u03b5\u03b9\u03bc\u03b9) — 'I am'.
+"""
diff --git a/packs/en/corpora/manifest.yaml b/packs/en/corpora/manifest.yaml
new file mode 100644
index 00000000..9f89a1fd
--- /dev/null
+++ b/packs/en/corpora/manifest.yaml
@@ -0,0 +1,13 @@
+corpus_id: en-tier0
+status: draft
+
+sources: []
+
+license_review: pending
+hashes: []
+
+notes: |
+  Populate with vetted, deterministically segmentable source manifests.
+  Each source must include: title, origin, license, SHA-256 of raw file,
+  and structural segmentation strategy (e.g. paragraph-boundary, sentence-boundary).
+  No LLM-extracted content. Segmentation must be D0 or D1.
diff --git a/packs/en/orthography.yaml b/packs/en/orthography.yaml
new file mode 100644
index 00000000..a8115d44
--- /dev/null
+++ b/packs/en/orthography.yaml
@@ -0,0 +1,21 @@
+language: en
+script: Latin
+
+normalization:
+  unicode_normal_form: NFC
+  lowercase_for_lookup: true
+  strip_diacritics: false
+
+punctuation:
+  strategy: preserve
+  sentence_terminals: [".", "!", "?"]
+  quotation_marks: ["\u201c", "\u201d", "\u2018", "\u2019"]
+
+spacing:
+  strategy: whitespace-delimited
+  collapse_multiple_spaces: true
+
+special_forms:
+  contractions: expand
+  # "don't" -> "do not" before lemmatization
+  # "it's"  -> "it is" before lemmatization
diff --git a/packs/en/pack.toml b/packs/en/pack.toml
new file mode 100644
index 00000000..9d9847f3
--- /dev/null
+++ b/packs/en/pack.toml
@@ -0,0 +1,30 @@
+pack_id     = "en"
+name        = "English Base Pack"
+version     = "0.1.0"
+status      = "draft"
+script      = "Latin"
+default     = true
+
+[contracts]
+lemma_first          = true
+shared_field_target  = true
+local_readback       = true
+alignment_required   = true
+
+[gates]
+schema       = false
+lexical      = false
+morphology   = false
+lift         = false
+readback     = false
+determinism  = false
+alignment    = false
+coverage     = false
+
+[notes]
+role = """
+Default articulation base. English is chosen as the default for practical
+reasons (current primary working language) but the contract does not privilege
+it semantically. Any language could serve this role in a custom CORE instance.
+The default flag controls surface output, not field-layer priority.
+"""
diff --git a/packs/he/corpora/manifest.yaml b/packs/he/corpora/manifest.yaml
new file mode 100644
index 00000000..c3a9aadb
--- /dev/null
+++ b/packs/he/corpora/manifest.yaml
@@ -0,0 +1,15 @@
+corpus_id: he-tier0
+status: draft
+
+sources: []
+
+license_review: pending
+hashes: []
+
+notes: |
+  Populate with vetted Hebrew source manifests.
+  Priority tier-0 sources: Biblia Hebraica Stuttgartensia (BHS) or
+  Westminster Leningrad Codex (WLC) — both have well-established structural
+  segmentation (book/chapter/verse = D0 boundary markers).
+  Each source must include: title, edition, license, SHA-256, and
+  segmentation strategy. No LLM-extracted content.
diff --git a/packs/he/orthography.yaml b/packs/he/orthography.yaml
new file mode 100644
index 00000000..92fafd8e
--- /dev/null
+++ b/packs/he/orthography.yaml
@@ -0,0 +1,27 @@
+language: he
+script: Hebrew
+direction: rtl
+
+normalization:
+  unicode_normal_form: NFC
+  lowercase_for_lookup: false  # Hebrew has no case
+  strip_diacritics: configurable
+  # Nikkud (vowel pointing) is stripped by default for lemma lookup
+  # but preserved when semantically load-bearing (e.g. Qere/Ketiv distinction)
+  nikkud_policy: strip_for_lookup
+  cantillation_marks: strip
+
+punctuation:
+  strategy: preserve
+  paragraph_marker: "\u05c0"  # paseq
+  section_marker: "\u05c3"    # sof pasuq
+
+spacing:
+  strategy: whitespace-delimited
+  maqaf_handling: preserve
+  # maqaf (\u05be) connects words syntactically; preserve as a unit marker
+
+shoresh:
+  # Root-based morphology. Lemma lookup resolves to 3- or 4-letter root.
+  # Morphological analysis must identify shoresh before feature extraction.
+  extraction_required: true
diff --git a/packs/he/pack.toml b/packs/he/pack.toml
new file mode 100644
index 00000000..218c5285
--- /dev/null
+++ b/packs/he/pack.toml
@@ -0,0 +1,32 @@
+pack_id     = "he"
+name        = "Hebrew Depth Pack"
+version     = "0.1.0"
+status      = "draft"
+script      = "Hebrew"
+default     = false
+
+[contracts]
+lemma_first          = true
+shared_field_target  = true
+local_readback       = true
+alignment_required   = true
+
+[gates]
+schema       = false
+lexical      = false
+morphology   = false
+lift         = false
+readback     = false
+determinism  = false
+alignment    = false
+coverage     = false
+
+[notes]
+role = """
+Depth language pack 1. Hebrew was chosen because of the range of depth it
+brings to the vocabulary manifold: root-based morphology (shoresh), semantic
+families that carry multiple layers of meaning simultaneously, and the
+language in which the universe was spoken into existence (Genesis 1).
+Hebrew morphological stems (Qal, Niphal, Piel, Pual, Hiphil, Hophal,
+Hithpael) are semantically load-bearing and must be represented natively.
+"""