fluree docs

June 24, 2026 · View on GitHub

Search and read the Fluree documentation from the command line. The docs are embedded in the fluree binary, so lookups work offline and are version-exact for this build — they match the Fluree you're developing against, with no "latest"-vs-installed drift.

This is the human-facing CLI. The same corpus is exposed to IDE agents as the docs toolset of the MCP server — see fluree mcp.

Usage

fluree docs <COMMAND>

Subcommands

CommandDescription
searchRanked, section-level search
getPrint a page, or one heading section, as markdown
examplesExtract code examples for a topic
treePrint the documentation table of contents

No .fluree/ directory or running database is required — these work in any directory.

fluree docs search <QUERY> [--limit <N>] [--json]

Returns ranked hits at heading-section granularity — each with the page path, heading anchor, title, a snippet, and a relevance score (BM25 over titles, headings, and body, with stemming).

OptionDescription
--limit <N>Max hits (default: 10)
--jsonEmit JSON instead of human-readable text
fluree docs search "property paths"

fluree docs get

fluree docs get <PATH> [--anchor <ANCHOR>] [--json]

Prints a whole page as markdown, or just one section when --anchor is given. Use the path and anchor from a search hit.

OptionDescription
--anchor <ANCHOR>Return only the section with this heading anchor
--jsonEmit JSON ({path, title, anchor, content, version})
fluree docs get query/sparql.md --anchor property-paths

fluree docs examples

fluree docs examples <QUERY> [--lang <LANG>] [--limit <N>] [--json]

Extracts fenced code blocks from the sections most relevant to the query — often all you need to get the syntax right in one shot.

OptionDescription
--lang <LANG>Only return code blocks in this language (e.g. json, sparql)
--limit <N>Max examples (default: 10)
--jsonEmit JSON
fluree docs examples "insert transaction" --lang json

fluree docs tree

fluree docs tree [--json]

Print the documentation table of contents — the curated SUMMARY.md hierarchy of titles and page paths. Use it to orient (see what topics exist) and grab a page path to feed get, rather than inferring structure from search results.

OptionDescription
--jsonEmit JSON ({nodes: [{title, path, children}], version}) instead of an indented tree
fluree docs tree

Only pages listed in SUMMARY.md appear; a markdown file not in the TOC is still searchable and retrievable by path, but won't show in the tree.

Docs for IDE agents (MCP)

The same lookup is exposed to IDE agents as the docs toolset of the unified Fluree MCP server — there is no separate docs server. It is read-only over static, embedded content (safe to auto-allow), and surfaces four tools:

ToolDescription
docs_searchRanked, section-level hits (query, limit?)
docs_getA page or one heading section (with its subtree) as markdown (path, anchor?)
docs_examplesCode examples for a topic (query, lang?, limit?)
docs_treeThe documentation table of contents, for browse/orientation (no args)

Every result carries a version field matching the binary, so the agent can trust it over training-data recall.

Register it with your IDE (docs only, no .fluree/ directory needed):

fluree mcp init --toolsets docs

See fluree mcp for the full command surface, manual config, and combining docs with the memory toolset.

See Also

  • mcp — the unified MCP server (fluree mcp init / serve / status) and its toolsets
  • Memory: IDE support matrix — per-IDE MCP config file paths and keys