Skip to content

Guides

Export: the CONTEXT folder, JSON and llms.txt

You can take the whole brain with you at any time: the export returns a project as a ZIP of markdown files with the links between pages, the messages, the images and the documents, or as JSON. Every page of the book also has a markdown twin, and each project an llms.txt.

The brain is yours, and leaving should never mean losing it. An export is a snapshot: it is complete at the moment you take it, and nothing syncs back. To change the context, write to the graph (in the book or over MCP); to refresh a copy, export again.

From the settings

In a project, open Settings › Brain › Project and choose Download export under Export. With full history adds every earlier version of every block. In a project that inherits from its organisation, a third button downloads the organisation's shared context: that is the export of the organisation's own project, because a project's export holds only its own blocks.

The export endpoint

curl -H "Authorization: Bearer $EPICCONTEXT_API_KEY" \
  -o project.zip \
  "https://www.epiccontext.com/api/v1/projects/$PROJECT_ID/export"

It needs an API key with the read scope. Three query parameters:

ParameterValuesDefault
formatzip or jsonzip
includeDocumentstrue or falsetrue
historytrue adds history.json: every earlier version of every block, archived ones included, with its reasonfalse

The settings buttons download the same ZIP, documents included.

What the ZIP holds

CONTEXT/
  README.md                 the entry point: what the project is and where things are
  AI-GUIDE.md               how the brain is shaped, for AI tools that read files
  .ai-guides/               one guide per section
  <section>/README.md       per section, with its guide
  <section>/…/<key>.md      one markdown file per block, with frontmatter
  decisions/adr-<n>-<slug>.md
  links.json                the typed links between blocks
  messages.json             the messages left on pages
  assets/                   the images the pages show
  history.json              with history=true only
DOCUMENTS/
  <folders>/<files>         the project's uploaded documents
  .manifest.json

Each block file carries its type, section, key, name, maturity and dates in YAML frontmatter, and its links to other blocks; the body is the same markdown the book renders, with its images pointing into assets/. Archived blocks are left out. links.json names each link's two blocks, its type, its note and its reason, and says whether the other end is in this export. AI-GUIDE.md is generated at export time, so never edit it by hand.

The JSON export

format=json returns the same files as data, links.json, messages.json and (on request) history.json among them, for a pipeline that writes its own copy. The images stay links to their files:

{
  "project": { "id": "…", "name": "…", "slug": "…" },
  "files": [{ "path": "users/personas/primary-persona.md", "content": "…", "updatedAt": "…" }],
  "totalFiles": 42,
  "deletions": [],
  "documents": [],
  "totalDocuments": 0
}

deletions lists the blocks archived in the last thirty days, so a copy can drop them too.

Without an export

  • One page as markdown: add .md to any page's address in the book. You get its body and its open messages.
  • The book's index for AI tools: /<organisation>/<project>/book/llms.txt lists the written pages with links to their markdown twins.
  • Everything in one document, over MCP: get_book returns the whole project as markdown, or one chapter.

The book, its twins and llms.txt are for members of the organisation only.

Is there a CLI or a sync?

No. There is no npm package and no push or pull. The connector and the plugin come from the server, and the CONTEXT folder exists only as an export. If an older guide told you to run a sync command, use the connector instead.

Next

Last updated: