From theseus-pro
Use when tasks involve the Theseus/Index-API intelligence stack: evidence intake, claim extraction, contradiction and tension analysis, learned scoring, GNN and temporal graph memory, KGE RotatE, SBERT enrichment, GL-Fusion, model-swarm routing, compound learning, IQ measurement, self-organization, symbolic/counterfactual reasoning, or two-mode Railway vs local deployment decisions. Loads command routing, invariants, and source-first execution rules for the full engine.
How this skill is triggered — by the user, by Claude, or both
Slash command
/theseus-pro:intelligence-engineThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Read live source before writing code. Read `refs/` for library internals.
Read live source before writing code. Read refs/ for library internals.
Read the Index-API codebase for application patterns. Do not rely on
training data for either. Treat specs as intent; reconcile with current
implementation before editing.
/reason — text -> claims -> tensions -> models (NLP, NLI, KR).
Load agents: claim-analysis, nlp-pipeline, knowledge-representation,
information-retrieval, probabilistic-reasoning.
/graph — objects -> structure -> self-organization (graph, causal,
GNN, temporal memory).
Load agents: graph-theory, causal-inference, self-organization,
knowledge-representation, probabilistic-reasoning, graph-neural-networks,
temporal-graph-memory, systems-theory.
/train — features -> models -> evaluation -> adaptation (ML, RL,
evolution, enrichment pipeline).
Load agents: learned-scoring, training-pipeline, graph-neural-networks,
temporal-graph-memory, language-model-training, multimodal-networks,
reinforcement-learning, evolutionary-optimization, domain-specialization.
/architect — system design, feedback loops, pipeline optimization.
Load agents: software-architecture, self-organization, program-synthesis,
systems-theory, evolutionary-optimization, domain-specialization.
/simulate — hypotheses, debate, counterfactuals, belief revision.
Load agents: claim-analysis, causal-inference, probabilistic-reasoning,
multi-agent-reasoning, symbolic-reasoning, counterfactual-simulation,
language-model-training, systems-theory, temporal-graph-memory.
/measure — IQ tracking across 7 axes, benchmarking, leverage analysis.
Load agents: information-retrieval, graph-theory, self-organization,
learned-scoring, reinforcement-learning, systems-theory, domain-specialization.
/learn — compound learning lifecycle.
Save session log, run confidence updates, review auto-captured claims,
resolve tensions, and process stale/low-confidence attention items.
knowledge/claims.jsonl as live operational memory for reusable
practices.knowledge/session_log/ and run /learn after
meaningful implementation or debugging sessions.Edge.reason is plain-English, human-readable.Edge uses from_object / to_object (not source/target).retrospective_notes)._generate_sha().engine_config controls pass behavior.is_deleted=True).compose_engine is stateless (text-in, objects-out, no DB writes).engine.py is stateful (object-in, edges + nodes out).Use these pattern files when implementing the above:
patterns/PATTERNS-epignn.mdpatterns/PATTERNS-kge-rotate.mdpatterns/PATTERNS-sbert-enrichment.mdpatterns/PATTERNS-gl-fusion-three-stream.mdpatterns/PATTERNS-compound-learning.mdpatterns/PATTERNS-model-swarm.mdpatterns/PATTERNS-network-effects.md| Level | Name | Status |
|---|---|---|
| 1 | Tool-Based Intelligence | Shipped |
| 2 | Learned Connection Scoring | Building |
| 3 | Hypothesis Generation | Designed |
| 4 | Emergent Ontology | Designed |
| 5 | Self-Modifying Pipeline | Designed |
| 6 | Multi-Agent Epistemic Reasoning | Designed |
| 7 | Counterfactual Simulation | Designed |
| 8 | Creative Hypothesis Generation | Designed |
Discovery (0.20), Organization (0.15), Tension (0.15), Lineage (0.10),
Retrieval (0.15), Ingestion (0.10), Learning (0.15).
Current composite: check live via theseus_measure_iq MCP tool.
patterns/PATTERNS-*.md for implementation recipes.product/*.md for specs, roadmap, and design docs.examples/{train,simulate,measure,reason,graph}/*.py for code.agents/*.md for domain-specific expertise (24 agents, 3 tiers).knowledge/*.jsonl and knowledge/manifest.json.Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub travis-gilbert/claude-marketplace --plugin theseus-pro