cairnv1.2.2

Development teams ingest their repositories into cairn to enable hybrid FTS5 + vector search across code, docs, and design notes. Engineers query natural language questions without leaving the terminal or IDE, getting ranked chunks plus AST-based call graphs. This replaces fragile keyword grep workflows with a curated, local-first index that never sends code to external services.

Legal, finance, and defense organizations use cairn with CAIRN_OFFLINE=1 and pre-cached GGUF models to index audit notes, policies, and PDFs entirely offline. The path allowlist and size caps enforce ingestion safety for multi-user or shared deployments. Retrieval stays on localhost with no telemetry or accounts, satisfying strict data residency requirements.

Teams building agents over MCP expose cairn's search, graph, ask, path, and tags surfaces to their LLM tools, letting models query and maintain a curated local knowledge base. Disable-model-invocation ensures users explicitly control what enters the index, preventing accidental exfiltration. This enables grounded RAG without managing a cloud vector database.

Blockchain developers ingest SDK source and on-chain program code as separate sources, then use `cairn link` to resolve cross-source name references between them. The graph surfaces calls, depends_on, and mitigates edges across the SDK-program boundary, and tag filters isolate concepts like 'invariant' or 'attack'. This gives audit and protocol teams fast, traceable retrieval over distributed codebases.

Researchers, writers, and analysts collect web pages, PDFs, and raw text notes into a single sqlite file, then query across the whole corpus with hybrid search and tag-filtered asks. The local-first design means no cloud sync, no accounts, and full ownership of the index. It works well as a long-lived second brain that survives tooling churn.

The core cairn index stays MIT-licensed and free for individual developers, while advanced path allowlist enforcement, audit logging, and multi-tenant index isolation are offered in a paid enterprise edition. This mirrors successful open-core database models by monetizing compliance and scale needs rather than basic retrieval. The safety gates introduced in v1.2 signal that enterprise-grade ingestion controls are a natural paid tier.

Offer consulting and deployment services for organizations that need cairn running inside air-gapped or regulated environments, including pre-cached GGUF model delivery, SHA256 verification workflows, and allowlist configuration hardening. Customers pay for one-time setup plus ongoing maintenance retainers. This monetizes the operational complexity that the SKILL.md explicitly acknowledges around offline and compliance-sensitive deployments.

Sell pre-built, verified cairn index packs for common domains such as security standards, framework documentation, or regulatory reference sets, delivered as sqlite files with tag vocabularies already applied. Buyers get instant grounding without building their own ingestion pipeline. Revenue comes from one-time pack purchases plus subscription access to updated packs.