TheseusCodex

Method composition

A method is rarely atomic. Theseus's published methods rest on each other: an extractor feeds a coherence judge, which feeds a synthesis step. This map shows the public-visible portion of that dependency graph. The internal version of this graph carries drift state and failure-mode flags; only the structure is published here.

classify_claim_typecontradiction_geometrycontradiction_probedecompose_voiceexternal_claim_matchextract_claimsextract_methodologyextract_predictionmethod_candidate_extractornli_scorersix_layer_coherencesuggest_researchsynthesize_conclusion

Methods and their dependencies

The same dependency structure drawn above, as a card per method — this is the primary view on a phone, where the radial map is too dense to read.

  • classify_claim_typev1.0.0
    Classifies a claim into discourse categories (METHODOLOGICAL, SUBSTANTIVE, etc.).
    composes:
  • contradiction_geometryv1.0.0
    Detects contradiction via Hoyer sparsity of embedding difference vectors.
    composes:
  • contradiction_probev1.0.0
    Predicts the embedding-space neighborhood where a new proposition's logical contradiction should lie, then surfaces nearby existing propositions as unconfirmed candidates.
    composes: contradiction_geometry
  • decompose_voicev1.0.0
    Decomposes a founder's voice into an intellectual profile with orientation scores.
    composes:
  • external_claim_matchv1.0.0
    Ingests external literature and matches claims against internal positions.
    composes:
  • extract_claimsv1.0.0
    Extracts atomic truth-apt claims from a text chunk using an LLM.
    composes:
  • extract_methodologyv1.0.0
    Extracts portable methodology profiles from a source text.
    composes:
  • extract_predictionv1.0.0
    Extracts falsifiable world predictions from a single claim using an LLM.
    composes:
  • method_candidate_extractorv1.0.0
    Scans ingested artifacts for passages describing a methodology and extracts structured method candidates via regex + LLM.
    composes:
  • nli_scorerv1.0.0
    NLI cross-encoder scorer for claim pair coherence using DeBERTa.
    composes:
  • six_layer_coherencev1.0.0
    Six-layer coherence aggregation with 4/6 majority voting.
    composes: nli_scorer
  • suggest_researchv1.0.0
    Generates research topics, empirical anchors, and reading lists after a discussion.
    composes:
  • synthesize_conclusionv1.0.0
    Registers a substantive conclusion and returns method calibration feedback.
    composes: extract_claims · nli_scorer · six_layer_coherence

Snapshot regenerated on each deploy. The same depends-on declarations drive both this page and the internal composition view.