okf-skills

Generic Model Context Protocol (MCP) server that discovers installed okf-* skill binaries and exposes their produce/ingest commands as MCP tools for any MCP-capable harness (Claude Code, Gemini CLI, etc.). Use when wiring OKF Skills into an agent over MCP instead of invoking each connector binary directly.

Google Cloud BigQuery connector that produces and ingests Open Knowledge Format (OKF) bundles from dataset schemas, table/field descriptions, and metadata. Use when documenting or cataloging BigQuery datasets, extracting schema metadata into OKF, or syncing descriptions back to BigQuery tables and fields.

CSV connector that produces and ingests Open Knowledge Format (OKF) bundles from a directory of CSV files. Infers each file's column schema (integer/number/boolean/date/string) by sampling rows, optionally embeds a per-column data profile and sample rows, and syncs descriptions back to a .okf-metadata.yaml sidecar. Use when documenting or cataloging a folder of CSV/flat files, or capturing their inferred schema and stats as an OKF bundle. CGO-free, pure Go.

Local filesystem connector that produces and ingests Open Knowledge Format (OKF) bundles documenting directory trees and file metadata, honoring .okfignore and .okf-metadata.yaml. Use when documenting or cataloging a local folder structure, or capturing filesystem metadata as an OKF bundle.

Git repository connector that produces and ingests Open Knowledge Format (OKF) bundles documenting repository file trees plus commit-history metadata (times, authors, messages), honoring .gitignore and .okfignore. Use when documenting or cataloging a Git repository as an OKF bundle.

GraphQL connector that produces and ingests Open Knowledge Format (OKF) bundles from a GraphQL SDL document. Parses the schema into one concept per user-defined type (object/input/interface/enum/union) and per root operation (query/mutation/subscription field), links fields and operations to the types they reference as native relationship edges, and syncs descriptions back to a .okf-metadata.yaml sidecar. Use when documenting or cataloging a GraphQL API from its schema, with no live server. Pure Go.

Deterministic validator for Open Knowledge Format (OKF) bundles. Checks spec conformance (root index.md carries only okf_version, subdirectory index.md files carry no frontmatter, every concept has a non-empty type) plus enrichment coverage (placeholder descriptions, broken cross-links, orphans), and gates CI via its exit code. Use to validate an OKF bundle before publishing or ingestion, or to fail a build when a bundle regresses. Complements skills-ref validate, which only checks SKILL.md frontmatter.

MongoDB connector that produces and ingests Open Knowledge Format (OKF) bundles by sampling documents to infer each collection's schema. Emits one concept per collection with a Name | Type | Presence table (top-level fields, with mixed types shown as a union and presence as the percentage of sampled documents). The connection URI is bound via the MONGODB_URI environment variable and stripped of credentials before being recorded. ingest verifies the bundle against the database and, with --sync, creates missing collections; descriptions stay in the bundle. Use when documenting or cataloging a MongoDB database. CGO-free pure Go.

MySQL connector that produces and ingests Open Knowledge Format (OKF) bundles from database schemas and table/column comments. Use when documenting or cataloging a MySQL database, extracting its schema and comments into OKF, or syncing descriptions back to MySQL via DDL.

OpenAPI/Swagger connector that produces and ingests Open Knowledge Format (OKF) bundles from an API contract. Parses an OpenAPI 3.x or Swagger 2.0 spec (JSON or YAML) into one concept per operation (API Endpoint) and per component schema (Schema), links endpoints to the schemas they reference as typed cross-links, and syncs descriptions back to a .okf-metadata.yaml sidecar. Use when documenting or cataloging an HTTP API from its spec, with no live server or credentials. Pure Go.

PostgreSQL connector that produces and ingests Open Knowledge Format (OKF) bundles from database schemas and table/column descriptions. Use when documenting or cataloging a PostgreSQL database, extracting its schema and comments into OKF, or syncing descriptions back with COMMENT ON statements.

SQLite connector that produces and ingests Open Knowledge Format (OKF) bundles from database schemas and comments, CGO-free via modernc.org/sqlite. Use when documenting or cataloging a SQLite database, extracting its schema into OKF, or comparing/validating a database against an OKF bundle.

Visualizer that renders an Open Knowledge Format (OKF) bundle into a single self-contained interactive index.html — a three-pane explorer (tree/filters, a Cytoscape graph with switchable layouts, and a rendered reader) written next to index.md. Use when you want a human-browsable, graph view of an OKF bundle, or to generate a shareable HTML catalog from any connector's bundle.

Guidance for an AI agent to enrich an Open Knowledge Format (OKF) bundle with high-quality concept descriptions using its own LLM — grounded in the bundle's schema, data profile, and samples — then optionally sync them back to the source. Use when an OKF bundle has missing, weak, or low-quality descriptions that should be improved before publishing or ingestion. Instructions-only; no binary required.

Guidance and the bundled OKF specification for building a new Open Knowledge Format (OKF) producer/connector skill in this project — the architectural principles, the okf-go library contract, the produce/ingest/schema command surface, the secret-handling and sync conventions, and the exact registration steps. Use when creating a new OKF connector or producer for a data source the project does not yet cover (e.g. MongoDB, Redis, Kafka, CSV, an HTTP API), scaffolding an okf-* skill, or extending the project to a new source. Instructions-only; no binary required.

Guidance for AI agents on how to parse, traverse, and query Open Knowledge Format (OKF) bundles efficiently — minimizing token usage and avoiding slow recursive directory walks. Use when reading, navigating, analyzing, or answering questions about an existing OKF bundle. Instructions-only; no binary required.