Developer docs

Agent context pack: every approved doc in one call

One request returns every Current (approved) document in scope as Markdown: an index, then one section per document with its path, version, approvers, publish date and link. Drafts and open proposals never appear. Team plan and up. Team+

REST

bash

curl -H "Authorization: Bearer $VERSODOCS_API_KEY" \n  "https://versodocs.com/api/v1/orgs/acme/context-pack?folder=eng/specs&max_chars=150000" > CONTEXT.md

Query parameters

Name	Type	Description
`workspace`	`string`	Workspace slug: only that workspace.
`folder`	`string`	`<workspace>/<folder>/…`: that folder and everything below it.
`format`	`"md" \| "json"`	Markdown (default) or JSON with `documents`, `index`, `omitted`, `truncated`.
`max_chars`	`integer`	Character budget, default 200,000, up to 2,000,000.

Budget: documents are added in path order; the one that crosses the budget is cut and the rest are listed in a closing Truncated notice. Narrow with workspace / folder or raise max_chars.
Size cap: at most 500 documents per pack; larger scopes report how many were left out.
Caching: the ETag changes only when a document in scope gets a new Current version, is added, moved or renamed. Send If-None-Match to get 304 without re-downloading.
Scope: the key’s workspaces, exactly like every other read (workspace-scoped keys see one workspace; a member’s one-click connection sees only workspaces that member can open).

MCP: get_context_pack

Connected agents (Claude Code, Cursor, Copilot…) call get_context_pack with optional workspace, folder and max_chars to load the specs before starting a task. See MCP.

Private llms.txt

GET /api/v1/orgs/{org}/llms.txt returns an llms.txt-style index of your Current documents grouped by workspace, each linking its Markdown through the API. Tools that read llms.txt fetch the linked documents with the same API key.