AI & ML Integration

Terp Network provides infrastructure and tooling for integrating AI and ML capabilities into blockchain workflows.

QMD — Query Markdown Documentation

QMD is a semantic code analysis tool that indexes CosmWasm contract source code and documentation into a searchable database. It powers:

• Smart contract discovery — find contracts by behavior, not just name
• Semantic code search — query code by meaning, not exact strings
• Documentation generation — auto-generate API docs from indexed contracts
• Agent integration — LLMs use QMD indexes to answer questions about Terp Network contracts

# Index a contract collection
qmd collection add my-contracts ./contracts/
qmd index my-contracts

# Semantic search
qmd query "contracts that handle IBC callbacks"

QMD + Trailmark

Trailmark extends QMD with AST-level code graph analysis, enabling:

• Dependency mapping — visualize contract dependency trees
• Entrypoint analysis — understand which entrypoints a contract exposes
• Pattern detection — find common implementation patterns across contracts

QMD workflow diagram scaffold — a flow showing source code → QMD index → semantic queries → LLM/developer consumption. Include Trailmark AST analysis as a parallel path.

LLM Integration Points

Terp Network documentation and tooling are designed to work with LLM-based coding assistants:

llms.txt

Terp Network provides an llms.txt file that LLMs can ingest for context-aware assistance. This file includes:

• Core documentation pages in LLM-friendly format
• Code snippets and examples
• Import paths and module references
• Common patterns and best practices

Contract Development Assistance

LLMs can assist with CosmWasm contract development using QMD-indexed context:

• Generate contract scaffolding
• Explain IBC callback patterns
• Suggest fee grant implementations
• Debug compilation errors

// Example: LLM-generated IBC callback pattern
#[entry_point]
pub fn ibc_source_callback(deps: DepsMut, env: Env, msg: IbcSourceCallbackMsg) -> StdResult<Response> {
    // Validate the callback came from the expected channel
    // Process the acknowledgement result
    // Execute any conditional logic
}

Semantic Code Indexing

The Terp Network ecosystem uses semantic indexing to make codebases searchable by meaning rather than keywords.

Indexing architecture diagram scaffold — a system diagram showing how source code flows through QMD indexing, semantic search, and LLM consumption. Include collections, vector embeddings, and query interfaces.

Use Cases

1. Developer onboarding — new developers query the codebase semantically instead of grepping
2. Audit assistance — auditors traverse code graphs to find security-relevant paths
3. Automated documentation — generated docs stay in sync with indexed code
4. Agent tooling — autonomous agents use QMD indexes to answer user questions about specific contracts

Further Reading

• QMD setup guide — installation and configuration
• Semantic indexing docs — advanced usage
• LLM context configuration — machine-readable doc index