Skip to content

docs: add Agent Pack documentation#12056

Closed
julian-risch wants to merge 1 commit into
mainfrom
docs/add-agent-pack-docs
Closed

docs: add Agent Pack documentation#12056
julian-risch wants to merge 1 commit into
mainfrom
docs/add-agent-pack-docs

Conversation

@julian-risch

@julian-risch julian-risch commented Jul 17, 2026

Copy link
Copy Markdown
Member

Related Issues

Proposed Changes:

Adds documentation for the agent_pack integration (agent-pack-haystack), which currently isn't covered by the Haystack docs at all.

New "Agent Pack" category under Haystack Concepts → Agents (sibling of Multi-Agent Systems):

  • docs/concepts/agents/agent-pack.mdx — landing page: what the pack is, installation, overview table of the agents, shared usage pattern, and when to use the pack vs. building your own Agent.
  • docs/concepts/agents/agent-pack/deep-research-agent.mdxcreate_deep_research_agent: quick start, outputs, configuration (swappable LLMs + budgets), and a "How it works" architecture deep dive (Scope→Research→Write hooks, orchestrator/sub-researcher, context isolation & compression).
  • docs/concepts/agents/agent-pack/advanced-rag-agent.mdxcreate_advanced_rag_agent: quick start, outputs & citations, configuration, retrieval-pipeline example, and a deep dive (tool loop, DocumentStoreToolset reuse, filter grammar, supported document stores).

Each agent page is two-level: high-level factory-function usage on top for users who just want to import and run an agent, and lower-level architecture details below for users who want to copy the patterns.

Also adds a "Ready-made agents" bullet to docs/concepts/agents.mdx and registers the pages in sidebars.js. Pages target the next docs version only (no versioned_docs mirror), since agent-pack-haystack is not released on PyPI yet.

How did you test it?

  • npm run build in docs-website passes (config throws on broken links/anchors, so all cross-references are validated).
  • Verified in the build output that the three pages render at /docs/next/agent-pack, /docs/next/deep-research-agent, /docs/next/advanced-rag-agent, appear in the sidebar, and the ASCII architecture diagrams render in monospace.
  • Code snippets and parameter defaults are taken from the package READMEs and factory docstrings in haystack-core-integrations.

Notes for the reviewer

  • The API reference rows link to https://docs.haystack.deepset.ai/reference/integrations-agent-pack as a full external URL: the reference page is CI-generated on the first agent_pack release tag, so a root-relative link would fail the build today. Each occurrence carries a TODO comment to switch after release.
  • The landing page has a :::note with the git-install command; it should be removed once agent-pack-haystack is on PyPI. Draft until the release timing is clear.

Checklist

  • I have read the contributors guidelines and the code of conduct.
  • I have updated the related issue with new insights and changes.
  • I have added unit tests and updated the docstrings. (docs-only)
  • I've used one of the conventional commit types for my PR title: docs:.
  • I have documented my code.
  • I have added a release note file (docs-website change, no release note needed)
  • I have run pre-commit hooks and fixed any issue.

🤖 Generated with Claude Code

Document the agent_pack integration (agent-pack-haystack) under
Haystack Concepts > Agents: a landing page plus one page per agent
(Deep Research Agent, Advanced RAG Agent), each with factory-function
quick start on top and a README-style architecture deep dive below.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@vercel

vercel Bot commented Jul 17, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
haystack-docs Ready Ready Preview, Comment Jul 17, 2026 3:19pm

Request Review

Comment on lines +64 to +66
## When to Use Agent Pack vs. Building Your Own Agent

- **Use an Agent Pack agent** when your use case matches what it does — a cited report from web research, or question answering over a document store with rich metadata — and you want working prompts, tool wiring, and budgets out of the box.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@julian-risch @anakin87 I think having a section like this makes sense

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants