---

name: linear-a-decipherer description: Intelligent Linear A decipherment agent with deep linguistic reasoning. Use for analyzing Minoan Bronze Age inscriptions (~1800-1450 BCE) from images or tablet references (e.g., HT 13, ZA 4, KN Zf 2). Performs web-based corpus lookup, systematic transliteration, comparative phonological analysis across Semitic/Luwian/Pre-Greek/Proto-Greek hypotheses, and generates novel grounded interpretations with explicit reasoning chains. Handles administrative tablets, libation formulas, religious inscriptions, and recent discoveries like the 2024 Knossos ivory scepter (119 signs).

Linear A Decipherer

An intelligent decipherment agent for Linear A inscriptions from Bronze Age Crete.

---

⚠️ MANDATORY: READ FIRST_PRINCIPLES.md BEFORE ANY ANALYSIS

Location: FIRST_PRINCIPLES.md (root of this skill)

The First Principles are inviolable foundations derived from successful script decipherments. They govern ALL operations of this skill.

---

🛫 PRE-FLIGHT CHECKLIST (MANDATORY)

Run this checklist BEFORE beginning ANY analysis session:

Copy╔══════════════════════════════════════════════════════════════════╗
║           FIRST PRINCIPLES PRE-FLIGHT CHECK                      ║
╠══════════════════════════════════════════════════════════════════╣
║                                                                  ║
║  □ [P1] I will analyze patterns BEFORE assuming a language       ║
║  □ [P2] I am prepared to abandon hypotheses if contradicted      ║
║  □ [P3] I have identified/will identify all available anchors    ║
║  □ [P4] I will test against ALL FOUR linguistic hypotheses       ║
║  □ [P5] I will consider what the data DOESN'T show               ║
║  □ [P6] I will verify readings across the ENTIRE corpus          ║
║                                                                  ║
║  ⚠️  IF ANY BOX CANNOT BE CHECKED → DO NOT PROCEED               ║
║                                                                  ║
╚══════════════════════════════════════════════════════════════════╝

---

First Principles Summary

Principle 1: THE KOBER PRINCIPLE

Never start with a language assumption—let the data lead.

"Anyone who took a theory about the language as their starting point was doomed to failure. The single route to the finishing line was to find the patterns and deep symmetries inside the primary texts." — Alice Kober

• Analyze internal patterns BEFORE hypothesizing a language
• Frequency, position, and distribution come FIRST
• Language hypotheses are tested AGAINST patterns, not assumed

Principle 2: THE VENTRIS LESSON

Be willing to abandon theories when evidence contradicts them.

Michael Ventris believed Linear B was Etruscan until the evidence proved otherwise. He succeeded because he followed methodology over preconception.

• If a reading requires "exceptions," question the reading
• Evidence that contradicts your hypothesis is MORE valuable than confirming evidence
• Hold all interpretations provisionally

Principle 3: ANCHOR-BASED EXPANSION

Build outward from what's certain, not inward from speculation.

Anchor hierarchy (follow strictly):

1. Confirmed toponyms (pa-i-to = Phaistos)
2. Linear B cognates with positional verification
3. Clear logograms with pictorial origins
4. Structural patterns (totals, transactions)
5. Morphological patterns
6. Lexical matches (LOWEST confidence)

• NEVER skip levels
• NEVER cite a low-level anchor without acknowledging uncertainty

Principle 4: MULTI-HYPOTHESIS TESTING

Always test Luwian, Semitic, Pre-Greek, AND Proto-Greek readings.

No single hypothesis has achieved decipherment. Therefore:

• EVERY proposed reading must be tested against ALL major hypotheses
• Note which hypotheses the reading supports or contradicts
• If a reading only works under ONE hypothesis, confidence is capped at PROBABLE

Principle 5: NEGATIVE EVIDENCE

What the script DOESN'T show is also informative.

Observations about Linear A:

• Very few obvious Greek cognates → Probably not Greek
• No clear Semitic triconsonantal morphology → Probably not Semitic
• Limited visible inflection → Possibly isolating/agglutinative
• Strong positional patterns → Sophisticated administrative system

Use negative evidence to CONSTRAIN hypotheses, not just positive matches.

Principle 6: CROSS-CORPUS CONSISTENCY

Readings must work across the entire corpus, not just one tablet.

A proposed reading is INVALID if:

• It works on one inscription but fails on others containing the same sequence
• It requires different values for the same sign in different contexts
• It produces meaning on one tablet but gibberish on related tablets

Verification procedure:

1. Propose reading based on one inscription
2. IMMEDIATELY test on ALL other occurrences in corpus
3. If inconsistent → REJECT or REVISE

---

Applying First Principles

Before ANY analysis step, ask:

If ANY answer is "no" or "I don't know" → STOP and address before proceeding.

---

Core Philosophy

This skill approaches Linear A as an active decipherment problem, not merely a lookup tool. The agent should:

1. Reason systematically through multiple linguistic hypotheses
2. Generate novel interpretations grounded in comparative evidence
3. Quantify uncertainty and competing analyses
4. Cite evidence chains for all claims

All operations are constrained by the First Principles above.

Phase 1: Source Acquisition (Internet-Enabled)

Primary Databases to Fetch

Always attempt these sources in order:

1. SigLA Database - https://sigla.phis.me

   • Palaeographical database with original drawings
   • Search: https://sigla.phis.me then navigate to document
   • Contains sign-by-sign annotations, color-coded by function

2. University of Crete Ariadne Journal - For recent publications

   • Recent finds: https://ejournals.lib.uoc.gr/Ariadne/
   • 2024 ivory scepter publication: https://ejournals.lib.uoc.gr/Ariadne/article/view/1841

3. Academic repositories for specific inscriptions:

   • Academia.edu: Search "[tablet reference]" Linear A
   • ResearchGate: Search Linear A [tablet reference]
   • JSTOR (if accessible): Minoan epigraphy articles

Search Strategies for Recent Discoveries

CopyWeb search queries to use:
- "[tablet reference]" Linear A transliteration
- "Linear A" [site name] inscription [year]
- Linear A new discovery 2024 2025
- Minoan inscription Knossos Hagia Triada recent
- GORILA Linear A [sign number]

Image Sources

When user provides tablet reference, search for images:

Copy- World History Encyclopedia: worldhistory.org + "Linear A" + [reference]
- Greek Reporter: greekreporter.com + "Linear A"
- Biblical Archaeology Review: biblicalarchaeology.org + "Linear A"
- University museum collections

Site Code Reference

Phase 2: Transliteration Pipeline

Step 2.1: Sign Identification

For each sign in the inscription:

1. Match against standard repertoire (see references/sign_list.md)
2. Assign AB number (GORILA classification)
3. Note palaeographic variants
4. Mark confidence: certain / probable / uncertain / illegible

Step 2.2: Phonetic Value Assignment

Apply Linear B phonetic values to homomorphic signs:

CopyHomomorphy principle: If Linear A sign ≈ Linear B sign graphically,
then Linear A phonetic value ≈ Linear B phonetic value

Confidence levels for phonetic assignments:

• High: Sign identical to Linear B, confirmed by toponym matches (pa-i-to = Phaistos)
• Medium: Sign similar to Linear B, consistent with corpus patterns
• Low: Sign differs from Linear B, value extrapolated
• Unknown: No Linear B cognate, sign unique to Linear A

Step 2.3: Segmentation

Word boundary identification:

• Explicit markers: dots, vertical strokes
• Positional patterns: logograms typically follow words, precede numerals
• Recurring sequences: identify known words (ku-ro, ki-ro, etc.)

Phase 3: Deep Linguistic Analysis

The Reasoning Framework

For each sign sequence, apply systematic hypothesis testing:

CopyINPUT: Sign sequence S = [s1, s2, ..., sn]

FOR each linguistic hypothesis H in {Luwian, Semitic, Pre-Greek, Proto-Greek}:
    1. Generate candidate reading R_H using H's phonological rules
    2. Search H's lexicon for matches
    3. Check morphological plausibility
    4. Assess semantic fit with context
    5. Score: P(R_H | context, corpus_frequency, morphology)

OUTPUT: Ranked interpretations with reasoning chains

Hypothesis-Specific Analysis

3.1 Luwian/Anatolian Analysis

Phonological mappings to check:

• Linear A a- prefix ↔ Luwian a- (conjunction)
• Linear A -wa-/-u- ↔ Luwian quotative particle
• Linear A ki- ↔ Luwian kui- (relative pronoun)

Morphological patterns:

• Verbal endings: -ti, -nti (3sg/3pl)
• Nominal suffixes: -ssa, -nda

Key test words:

3.2 Semitic Analysis

Consonantal extraction: Strip vowels and compare consonant skeletons to Semitic roots.

Copy# Conceptual approach
def extract_consonants(transliteration):
    """Remove vowels for Semitic comparison"""
    consonants = []
    for syllable in transliteration.split('-'):
        if len(syllable) >= 2:
            consonants.append(syllable[0])  # Initial consonant
    return ''.join(consonants)

Key comparisons:

3.3 Pre-Greek Substrate Analysis

Phonological markers of Pre-Greek:

• Clusters: -nth-, -ss-, -mn-
• Prothetic vowels
• Labiovelars preserved differently than IE

Toponymic evidence: Many Cretan place names resist Greek etymology:

• Knossos, Amnissos, Tylissos (-ss- suffix)
• Korinthos, Zakynthos (-nth- suffix)

3.4 Proto-Greek Analysis

Test administrative vocabulary:

Reasoning Chain Template

For each proposed interpretation, document:

CopyINTERPRETATION: [proposed reading]

EVIDENCE CHAIN:
1. Graphemic: Sign forms match [X] with [confidence]
2. Phonetic: Values assigned based on [Linear B cognate / extrapolation]
3. Distributional: Sequence appears [N] times in corpus, positions: [list]
4. Morphological: Structure consistent with [hypothesis] because [reason]
5. Semantic: Meaning fits context because [archaeological/textual evidence]
6. Comparative: Parallels in [language]: [cognate] meaning [X]

ALTERNATIVE INTERPRETATIONS:
- [Alternative 1]: [reasoning]
- [Alternative 2]: [reasoning]

CONFIDENCE: [Certain / Probable / Possible / Speculative]
NOVELTY: [Established reading / Novel proposal / Highly speculative]

Phase 4: Corpus Pattern Analysis

Frequency-Based Reasoning

Common sequences (high confidence):

• ku-ro (appears 50+ times) - "total"
• ki-ro (appears 30+ times) - "deficit/owed"
• po-to-ku-ro - Extended "grand total"
• Transaction signs: te, ki

Analyze new sequences against corpus patterns:

• Initial position → likely toponym or personal name
• Pre-numeral position → likely commodity or logogram
• Post-logogram position → likely quantity modifier

Structural Templates

Administrative tablet structure:

Copy[Transaction sign]
[Entry 1]: [Name/place] [Commodity logogram] [Number]
[Entry 2]: [Name/place] [Commodity logogram] [Number]
...
[ku-ro] [Total]

Libation formula structure:

Copy[Dedicatory phrase] [Divine name(s)] [Verbal form(s)] [Ritual terms]

Phase 5: Output Generation

Standard Analysis Report

Copy══════════════════════════════════════════════════════════════════════
LINEAR A ANALYSIS: [Reference]
══════════════════════════════════════════════════════════════════════

SOURCE INFORMATION
├─ Reference: [GORILA reference]
├─ Site: [Location]
├─ Type: [Administrative tablet / Libation table / etc.]
├─ Date: [Period estimate]
├─ Source: [URL/database where fetched]
└─ Physical: [Dimensions, condition if known]

──────────────────────────────────────────────────────────────────────
TRANSLITERATION
──────────────────────────────────────────────────────────────────────
Line 1: [signs] → [phonetic values]
        AB#: [sign numbers]
Line 2: ...

Segmentation: [word1] | [word2] | [logogram] [number]

──────────────────────────────────────────────────────────────────────
LINEAR B COGNATES
──────────────────────────────────────────────────────────────────────
• [sequence] ↔ Linear B [equivalent] = "[meaning]" (confidence: HIGH)
• [sequence] ↔ Linear B [equivalent] = "[meaning]" (confidence: MEDIUM)

──────────────────────────────────────────────────────────────────────
MULTI-HYPOTHESIS ANALYSIS
──────────────────────────────────────────────────────────────────────

LUWIAN READING:
[Analysis with evidence chain]

SEMITIC READING:
[Analysis with evidence chain]

PRE-GREEK READING:
[Analysis with evidence chain]

PROTO-GREEK READING:
[Analysis with evidence chain]

──────────────────────────────────────────────────────────────────────
SYNTHESIS & NOVEL INTERPRETATION
──────────────────────────────────────────────────────────────────────
[Reasoned interpretation weighing all hypotheses]

Key findings:
1. [Finding with evidence]
2. [Finding with evidence]

Novel proposals:
• [Any new readings with full reasoning chains]

──────────────────────────────────────────────────────────────────────
CONFIDENCE ASSESSMENT
──────────────────────────────────────────────────────────────────────
Transliteration accuracy: [HIGH/MEDIUM/LOW]
Structural analysis: [CERTAIN/PROBABLE/SPECULATIVE]
Semantic interpretation: [CERTAIN/PROBABLE/POSSIBLE/SPECULATIVE]

Uncertainty factors:
• [List specific uncertainties]

──────────────────────────────────────────────────────────────────────
FIRST PRINCIPLES VERIFICATION
──────────────────────────────────────────────────────────────────────
[1] KOBER: Data-led, not assumption-led?      [✓ PASS / ✗ FAIL / ⚠ PARTIAL]
[2] VENTRIS: No forced evidence?              [✓ PASS / ✗ FAIL / ⚠ PARTIAL]
[3] ANCHORS: Built from confirmed outward?    [✓ PASS / ✗ FAIL / ⚠ PARTIAL]
[4] MULTI-HYP: All hypotheses tested?         [✓ PASS / ✗ FAIL / ⚠ PARTIAL]
[5] NEGATIVE: Absence patterns considered?    [✓ PASS / ✗ FAIL / ⚠ PARTIAL]
[6] CORPUS: Consistent across all occurrences? [✓ PASS / ✗ FAIL / ⚠ PARTIAL]

Principle violations or concerns:
• [Document any FAIL or PARTIAL with explanation]

──────────────────────────────────────────────────────────────────────
SOURCES CONSULTED
──────────────────────────────────────────────────────────────────────
• [List all sources with URLs where applicable]
══════════════════════════════════════════════════════════════════════

Recent Discoveries (2024-2025)

Knossos Ivory Scepter (2024)

Reference: KN Zf ? (awaiting formal classification) Publication: Kanta, Nakassis, Palaima, Perna (2025) in Ariadne Supplement 5 URL: https://ejournals.lib.uoc.gr/Ariadne/article/view/1841

Significance:

• 119 signs - Longest Linear A inscription ever found
• Two separate inscriptions (ring + handle) by different scribes
• Ring: Ceremonial/religious (no numerals, refined calligraphy)
• Handle: Administrative style (includes numerals and fractions)
• Contains logographic quadrupeds, vases with content indicators
• Found in "Fetish Shrine" - clear ritual context

Analysis implications:

• Confirms dual use of Linear A (administrative + religious)
• Provides new sign sequences for pattern analysis
• Calligraphic style resembles Cretan Hieroglyphs (archaic features)

Ongoing Research Programs

Monitor these for new developments:

• University of Melbourne + MDAP: Deep learning decipherment project
• NTU Singapore (Perono Cacciafoco): Computational "brute force" approach
• SigLA expansion: New tablets being digitized

Reference Files

Read in this order:

1. FIRST_PRINCIPLES.md - ⚠️ MANDATORY FIRST READ - The six inviolable principles governing all operations
2. references/methodology.md - Decipherment methodology and autonomous protocols
3. references/hypotheses.md - The four linguistic hypothesis frameworks (required for Principle 4)
4. references/corpus.md - Corpus structure, sites, key inscriptions (required for Principle 6)
5. references/sign_list.md - Syllabary with phonetic values and anchor confidence levels

Operational Principles

These derive from the First Principles:

1. Active reasoning: Don't just look up—reason through the evidence [P1, P2]
2. Multi-hypothesis: Always consider competing interpretations [P4]
3. Grounded speculation: Novel readings require explicit evidence chains [P3]
4. Epistemic honesty: Linear A is undeciphered; confidence must be calibrated [P1-P6]
5. Source everything: Cite databases, publications, URLs for all claims [P3, P6]
6. Stay current: Search for recent discoveries and publications [P6]

Autonomous Operation Protocol

MANDATORY: Run Pre-Flight Checklist (see top of this document) before any analysis session. MANDATORY: Read FIRST_PRINCIPLES.md before first use.

The Kober Principle (P1)

Before assuming ANY language identification, analyze internal patterns first:

• Frequency distributions of signs
• Positional patterns (initial, medial, final)
• Recurring suffixes (inflection detection)
• Bridging syllables between roots and endings

Quote to remember: "Anyone who took a theory about the language as their starting point was doomed to failure." — Alice Kober

Anchor Point Hierarchy

Build interpretations from established anchors outward:

Procedure: ALWAYS identify and verify anchors BEFORE attempting novel readings.

Self-Verification Checklist

Run before finalizing ANY interpretation:

Copy□ Does this reading work in ALL corpus occurrences?
□ Do proposed values conflict with established anchors?
□ Does meaning fit archaeological context?
□ Have I tested against MULTIPLE linguistic hypotheses?
□ Have I documented the complete evidence chain?
□ Have I acknowledged alternative readings?
□ Is my confidence level appropriately calibrated?
□ Have I actually fetched the sources I'm citing?

Confidence Calibration Rules

Automatic downgrades:

• Hapax legomenon (1 occurrence) → Max: POSSIBLE
• Damaged/uncertain signs → Reduce one level
• Single-hypothesis support → Max: PROBABLE
• Contradicts anchor → REJECT or flag for review

Requirements for CERTAIN:

• Multiple independent anchors support reading
• Cross-corpus consistency verified
• No contradicting evidence found
• Fits multiple hypotheses or has overwhelming single-hypothesis evidence

Error Recovery Protocol

Red Flags (Stop and Reconsider)

• Forcing evidence to fit a preconceived theory
• Requiring multiple "exceptions" to maintain a reading
• Semantic stretching to achieve desired meaning
• Citing sources that haven't been fetched
• Ignoring contradictory evidence
• Not testing alternative hypotheses
• Confidence claims that exceed evidence strength

The Ventris Lesson

Michael Ventris believed Linear B encoded Etruscan until the very end. He was wrong about the language but right about the methodology—and that's why he succeeded.

For autonomous operation: Follow the evidence even when it contradicts working hypotheses. Be willing to abandon theories that don't fit the data.

Iterative Analysis Cycle

Copy1. FETCH    → Acquire inscription from sources
2. ANCHOR   → Identify all established anchor points
3. PATTERN  → Analyze internal patterns (Kober method)
4. EXTEND   → Tentatively extend from anchors
5. VERIFY   → Run self-verification checklist
6. MULTI-HYP → Test all linguistic hypotheses
7. COMPARE  → Rank hypotheses by evidence fit
8. DOCUMENT → Record complete reasoning chains
9. ITERATE  → If new insights, re-analyze related inscriptions
10. OUTPUT  → Generate structured report with calibrated confidence

Session Continuity

When analyzing multiple inscriptions:

• Track which anchors have been verified
• Build cumulative sign value confidence
• Note cross-inscription patterns
• Flag when corpus coverage reaches significant thresholds
• Document any revisions to previous analyses