Skip to content

docs: add v3.0 tool and model integration pages and fix stale references - #763

Open
Himanshu040604 wants to merge 3 commits into
mainfrom
himanshu/v3-tools-models-docs
Open

docs: add v3.0 tool and model integration pages and fix stale references#763
Himanshu040604 wants to merge 3 commits into
mainfrom
himanshu/v3-tools-models-docs

Conversation

@Himanshu040604

Copy link
Copy Markdown
Contributor

What

Nine new pages for v3.0 tool and model integrations, and fixes across ~50 existing pages that taught removed APIs or retired models. Everything describes agno v3.0.4, re-verified against v3.0.5 (all nine backing modules byte-identical; the three changed files elsewhere are additive).

New pages (each registered in the nav, with an overview card or provider-tree entries):

Page Location Introduced
Knowledge Management tools/toolkits/others/knowledge-management v3.0.4
AtomicMail tools/toolkits/social/atomic-mail v3.0.2
SuperGrok OAuth models/providers/native/xai/supergrok-oauth (both provider trees) v3.0.0
Finance tools/toolkits/others/finance v3.0.0
Ramp Router models/providers/gateways/ramp/overview (both trees, catalog + compatibility rows) v3.0.0
CodeMode tools/toolkits/local/code-mode v3.0.0
MiniMax video tools/toolkits/models/minimax v3.0.0
Serply tools/toolkits/search/serply v3.0.2
WaveSpeed tools/toolkits/others/wavespeed v3.0.2

Fixes, grouped (commit 1): Groq retired-model sweep across 11 hand-written pages incl. the Preview-only vision situation · Gemini default rows → gemini-3.7-flash · Cerebras retired-scout Warnings + gpt-oss-120b defaults · OpenAI widened request-param types on 5 tables · Anthropic >=0.77.0 floor + stale output_format notes · Sheets/BigQuery legacy enable_* alias Notes · Toolkit reference id row + positional-tools Warning · MultiMCPTools removed from 11 hand-written pages with real connection-failure semantics documented · DuckDuckGo, Seltz, BrightData, Brandfetch, Redis, Mistral, metrics-reference, @tool smaller fixes · docs changelog + migration-guide entries that shipped without them.

Why

v3.0 shipped ahead of its documentation. Several pages taught removed APIs (MultiMCPTools, creds_path, the Metrics alias) and models retired at their providers — the Groq default our docs recommended had been returning 404s since mid-August. The nine features above had no pages at all.

Verified

All 11 code fences on the new pages byte-compile, resolve their imports against the v3.0.4 tree, pass kwarg-level signature validation against the real constructor signatures, and contain no undefined names. None was executed against a live service. Every Badge version was established by git ls-tree across tags — three diverge from what the release notes imply (AtomicMail, Serply, and WaveSpeed shipped in v3.0.2, not where the notes place them). Every model id on touched pages was checked against its provider's current catalog and deprecation page on 2026-09-01. check_install_coverage.py passes (the two pre-existing offenders are untouched); docs.json diffs to exactly the intended 33 lines.

Known follow-ups in files this PR touches

Four stale ids remain in pages this PR edits, deferred because they need decisions rather than swaps:

  • groq/usage/reasoning-agent.mdx uses deepseek-r1-distill-llama-70b, retired Oct 2025. Not a straight swap: the page teaches a reasoning-model + response-model pair, and Groq's named successor is not a reasoning model. Needs a restructure or a model choice.
  • mistral/overview.mdx recommends open-mistral-nemo, retired Jul 2026 (successor: Ministral 3 8B), and ministral-14b-2512, which is misspelled — the model card reads ministral-3-14b-25-12. mistral-large-latest (the agno code default) is also now a provider-deprecated alias family.
  • The OpenAI responses overview has a gpt-5-mini fence, deprecated with a Dec 11 2026 shutdown (OpenAI names gpt-5.6-terra).

These land in a follow-up alongside the Groq streaming work below.

Not fixable here — the regen queue

91 generated pages carry source: frontmatter and still name removed symbols or retired ids. Hand edits to those are overwritten at the next regeneration, so the path runs through the agno repo's cookbooks, then examples_sync:

  • Broken (404ing models / removed classes): examples/models/groq/* (12), examples/models/cerebras-openai/* (5), the MultiMCPTools example family (9 incl. multiple-servers-allow-partial-failure, whose cookbook source was deleted — that one needs removal, not regen), two reasoning pages, one agent-os demo.
  • Stale but functional (old Gemini ids still serve): ~60 pages across examples/models/google/*, examples/basics/*, multimodal and team examples.
  • Pipeline config: scripts/examples_sync/generate.py and description-overrides.json still reference retired ids and MultiMCPTools.

The existing cookbook-fixes branch already unblocks the five cerebras-openai pages and the Groq image examples.

Findings for other people

Finding Owner / route
Four release-note errors (Seltz "legacy path removed"; Sheets "enable_* removed"; KMT "old flags ignored" — they raise TypeError; Toolkit id "used by AgentOS" — nothing reads it but repr()). Docs now state code truth release notes
Groq streaming has no reasoning-token handling — raw <think> leaks and streams end early with qwen models; the in-repo pattern is MiniMax's stateful inline-tag filter framework (libs/)
agno.tools.code.errors.ResultTooLarge is exported but never raised host-side; the raising class is a same-named kernel-side RuntimeError in the bridge bootstrap CodeMode author
WaveSpeed poll_interval=1.0 default vs the provider's stated 2-second minimum between result requests WaveSpeed toolkit author
Four cookbooks with no TEST_LOG entry (minimax_tools — also still on OpenAIChat —, wavespeed_tools, serply_tools, knowledge_management_tools) cookbook owners
zai-glm-4.7 has left Cerebras's public catalog but remains in the overview's Supported Models row docs follow-up
claude-sonnet-4-5-20250929 (the agno default) is Active but its retirement window opens Sept 29, 2026 watch
New v3.0.5 surface needs pages next pass: llmman provider, GandrTools, MCPTools headers=, EmbeddingError next docs pass
No vendor toolkit page links its model-provider page or vice versa, corpus-wide docs, someday

Decisions

Nav placements: AtomicMail → Social beside Gmail (a mailbox with an agent identity, not a delivery API like Resend/SES). RampRouter → gateway tier (the OpenRouter/Portkey shape: routes across vendors). MiniMax → the vendor-model toolkit category (a model vendor's own media API, the Gemini/Groq-toolkit shape). WaveSpeed → Additional Toolkits (a hosting platform serving other vendors' models, the fal/replicate shape). CodeMode → Local. SuperGrok → inside the existing xAI group after Usage (google precedent).

Deliberate omissions — flagged rather than guessed:

  • SuperGrok plan entitlement: which plans grant api:access is unpublished and public claims contradict each other. The page documents the requested scope, the endpoints, and the exact 403 the code substitutes, and links x.ai.
  • reasoning_replay: its source carries explicit assumption markers (models/xai/responses.py:89, :116-117); omitted rather than documenting an assumption.
  • The YFinance characterisation (unofficial / rate-limited / personal use, from the FinanceTools PR body): the corpus has zero precedent for editorializing a provider; the Finance page states requirements only and the three finance toolkits are presented as coexisting by design, none deprecated.
  • WaveSpeed's poll discrepancy: a code-owner nit, not a documentation fact — stating it on the page would editorialize a gap the docs cannot fix. It is in the findings table instead.
  • Lock-free multi-process token refresh (SuperGrok): the rotation grace window is a docstring-only server claim with no code constant; absent from the page.

Groq: retired llama-3.3-70b-versatile / llama-3.1-8b-instant replaced with
openai/gpt-oss-120b / openai/gpt-oss-20b across 11 hand-written pages, with
Groq's dated shutdown in a Warning and the Preview-only vision situation
stated; image agent moved to qwen/qwen3.6-27b. Gemini default rows to
gemini-3.7-flash. Cerebras overviews: dated scout deprecation, gpt-oss-120b
default rows, retired-model note on parallel_tool_calls. OpenAI: the four
widened request-param types on five param tables. Anthropic: anthropic>=0.77.0
floor with the output_config explanation; two stale output_format notes.
Sheets/BigQuery: legacy enable_* aliases documented as accepted. Toolkit
reference: id row (reserved) and the positional-tools Warning. MultiMCPTools
removed from 11 hand-written pages; multiple-servers rewritten around one
MCPTools per server with the real connection-failure semantics. Smaller:
DuckDuckGo alias removal, Seltz max_documents (legacy SDK path retained,
pin stays >=0.2.0), BrightData output_path, Brandfetch async_tools,
RedisDB -> RedisDb with the RedisVectorDb disambiguation, Mistral
mistralai>=2.0.0, metrics reference alias claim, @tool title/annotations
rows, one stray retired Groq id. Docs changelog and migration guide gain
the entries that shipped without them; the Toolkit id entries drop the
unsupported 'used by AgentOS' claim.
Nine new pages, written from tag-pinned code (v3.0.4) and registered in the
nav: KnowledgeManagementTools, AtomicMail, SuperGrok OAuth for xAIResponses,
FinanceTools, RampRouter, CodeMode, MiniMax video tools, Serply, and
WaveSpeed. The two provider pages register in both provider trees with
catalog and compatibility rows; the toolkit pages get overview cards and
cross-link Notes on their sibling pages. Every version Badge was established
by git ls-tree across tags; AtomicMail, Serply and WaveSpeed shipped in
v3.0.2, not where the release notes place them.
Copilot AI lite review requested due to automatic review settings September 1, 2026 12:16
@mintlify

mintlify Bot commented Sep 1, 2026

Copy link
Copy Markdown

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated (UTC)
agno 🟢 Ready View Preview Sep 1, 2026, 12:18 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.

Copilot AI left a comment

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.

Pull request overview

Adds v3.0 integration coverage (new toolkits + new provider page) and updates existing docs to stop recommending removed APIs and retired model IDs, aligning the docs set with Agno v3.0.x behavior.

Changes:

  • Added new integration pages (toolkits + providers) and linked them into navigation/card indexes.
  • Replaced removed symbols (notably MultiMCPTools and the Metrics alias) with v3.0 equivalents and updated behavioral notes.
  • Swept model/provider docs to swap retired defaults (Groq/Gemini/Cerebras) and widen OpenAI request-param type docs.

Reviewed changes

Copilot reviewed 66 out of 66 changed files in this pull request and generated 5 comments.

Show a summary per file
File Description
tools/toolkits/web-scrape/brightdata.mdx Updates BrightData function docs.
tools/toolkits/social/atomic-mail.mdx Adds AtomicMail toolkit page.
tools/toolkits/search/serply.mdx Adds Serply toolkit page.
tools/toolkits/search/seltz.mdx Updates Seltz param/alias behavior.
tools/toolkits/search/duckduckgo.mdx Marks removed DuckDuckGo aliases.
tools/toolkits/overview.mdx Adds cards for new toolkits.
tools/toolkits/others/yfinance.mdx Cross-links to Finance toolkit.
tools/toolkits/others/wavespeed.mdx Adds WaveSpeed toolkit page.
tools/toolkits/others/knowledge.mdx Cross-links to Knowledge Management.
tools/toolkits/others/knowledge-management.mdx Adds Knowledge Management toolkit page.
tools/toolkits/others/google-sheets.mdx Documents legacy enable_* aliases.
tools/toolkits/others/financial-datasets.mdx Cross-links to Finance toolkit.
tools/toolkits/others/finance.mdx Adds Finance toolkit page.
tools/toolkits/models/minimax.mdx Adds MiniMax toolkit page.
tools/toolkits/local/code-mode.mdx Adds CodeMode toolkit page.
tools/toolkits/database/google-bigquery.mdx Documents legacy enable_* aliases.
tools/mcp/transports/streamable_http.mdx Removes MultiMCPTools examples.
tools/mcp/transports/stdio.mdx Removes MultiMCPTools examples.
tools/mcp/transports/sse.mdx Removes MultiMCPTools examples.
tools/mcp/server-params.mdx Updates description to MCPTools-only.
tools/mcp/overview.mdx Updates MCPTools lifecycle notes.
tools/mcp/multiple-servers.mdx Reframes multi-server usage without MultiMCPTools.
tools/mcp/dynamic-headers.mdx Updates dynamic headers multi-server example.
reference/tools/toolkit.mdx Documents new Toolkit id param + warning.
reference/tools/decorator.mdx Documents @tool title + annotations.
reference/run/metrics.mdx Removes Metrics alias guidance.
reference/models/openai.mdx Widens OpenAI param typing docs.
reference/models/openai-like.mdx Widens OpenAILike param typing docs.
reference/models/open-responses.mdx Widens OpenResponses param typing docs.
reference/models/groq.mdx Updates Groq default model ID.
reference/models/gemini.mdx Updates Gemini default model ID.
reasoning/usage/tools/groq-reasoning-tools.mdx Updates Groq reasoning example model ID.
other/v3-migration.mdx Adds v3 migration notes for toolkits/models.
other/v3-changelog.mdx Adds changelog notes for v3 deltas.
models/providers/native/xai/supergrok-oauth.mdx Adds SuperGrok OAuth page for xAIResponses.
models/providers/native/openai/responses/overview.mdx Widens Responses param typing docs.
models/providers/native/openai/completion/overview.mdx Widens completion param typing docs.
models/providers/native/mistral/overview.mdx Pins mistralai>=2.0.0 guidance.
models/providers/native/google/overview.mdx Updates Gemini default model ID.
models/providers/native/google/gemini-interactions.mdx Updates GeminiInteractions default ID.
models/providers/native/anthropic/usage/structured-output.mdx Updates structured-output request shape notes.
models/providers/native/anthropic/overview.mdx Adds anthropic>=0.77.0 requirement notes.
models/providers/model-index.mdx Adds Ramp + xai-responses keys + card.
models/providers/gateways/ramp/overview.mdx Adds Ramp Router provider page.
models/providers/gateways/groq/usage/tool-use.mdx Updates Groq usage model ID.
models/providers/gateways/groq/usage/structured-output.mdx Updates Groq structured-output model ID.
models/providers/gateways/groq/usage/reasoning-agent.mdx Updates Groq reasoning-agent response model ID.
models/providers/gateways/groq/usage/metrics.mdx Updates Groq metrics example model ID.
models/providers/gateways/groq/usage/knowledge.mdx Updates Groq knowledge example model ID.
models/providers/gateways/groq/usage/image-agent.mdx Updates Groq vision guidance + model ID.
models/providers/gateways/groq/usage/deep-knowledge.mdx Updates Groq deep-knowledge model ID.
models/providers/gateways/groq/usage/basic.mdx Updates Groq basic example model ID.
models/providers/gateways/groq/usage/basic-stream.mdx Updates Groq streaming example model ID.
models/providers/gateways/groq/overview.mdx Updates Groq defaults + multimodal notes.
models/providers/gateways/cerebras/overview.mdx Updates Cerebras deprecation/default notes.
models/providers/gateways/cerebras-openai/overview.mdx Updates CerebrasOpenAI deprecation/default notes.
models/compatibility.mdx Adds RampRouter + xAIResponses rows.
knowledge/vector-stores/redis/usage/redis-db.mdx Renames RedisDB -> RedisDb usage.
knowledge/vector-stores/redis/usage/async-redis-db.mdx Renames RedisDB -> RedisDb async usage.
knowledge/vector-stores/redis/overview.mdx Updates Redis vector store naming notes.
faq/import-errors.mdx Removes MultiMCPTools from import errors.
examples/tools/mcp/streamable-http-transport/overview.mdx Updates example overview wording.
examples/tools/mcp/sse-transport/overview.mdx Updates example overview wording.
examples/tools/mcp/overview.mdx Updates MCP examples index wording.
examples/tools/brandfetch-tools.mdx Removes Brandfetch async_tools arg.
docs.json Registers new pages in nav.
Suppressed comments (1)

examples/tools/mcp/overview.mdx:39

  • These rows claim the linked transport examples use only MCPTools, but the linked client pages still include MultiMCPTools snippets. Adjust the descriptions (or update/regenerate the client pages) so the overview matches the linked content.
| [Sse Transport](/examples/tools/mcp/sse-transport/overview) | Connect agents to an MCP server over SSE transport using MCPTools. |
| [Streamable Http Transport](/examples/tools/mcp/streamable-http-transport/overview) | Connect an Agno agent to a Streamable HTTP MCP server using MCPTools. |

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

| -------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `scrape_as_markdown` | Scrapes a webpage and returns content in Markdown format. Parameters: `url` (str) - URL to scrape. |
| `get_screenshot` | Captures a screenshot of a webpage and adds it as an image artifact. Parameters: `url` (str) - URL to screenshot, `output_path` (str, optional) - Unused, kept for compatibility. |
| `get_screenshot` | Captures a screenshot of a webpage and adds it as an image artifact. Parameters: `url` (str) - URL to screenshot. |

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Checked: no trailing whitespace on this line or any row of the table (a /[[:space:]]$/ scan over rows 71-76 finds none), and the closing pipes were never aligned here - the neighbouring rows run 212-254 characters. Pipe tables render identically either way, so there is nothing to even out. Leaving as-is.

| [Groq + MCP = Lightning Fast Agents](/examples/tools/mcp/groq-mcp) | Create a high-performance filesystem agent by combining Groq's fast LLM inference with the Model Context Protocol (MCP). |
| [Include Exclude Tools](/examples/tools/mcp/include-exclude-tools) | Filter which MCP server tools an agent can use with include_tools and exclude_tools. |
| [Multiple Servers](/examples/tools/mcp/multiple-servers) | Connect one agent to Airbnb and Brave Search MCP servers at once using MultiMCPTools. |
| [Multiple Servers](/examples/tools/mcp/multiple-servers) | Connect one agent to Airbnb and Brave Search MCP servers at once using one MCPTools instance per server. |

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Deliberate, and declared in the PR body ('Not fixable here - the regen queue'): the linked example pages carry source: frontmatter, so hand edits are silently overwritten at the next regeneration - the fix path runs through the agno repo's cookbooks, then examples_sync. The hand-written index cannot keep describing MultiMCPTools, which is removed in v3.0 and raises ImportError; these rows describe the examples as they regenerate, and the mismatch closes on the next sync.

Comment thread tools/toolkits/social/atomic-mail.mdx Outdated
| Example | Description |
|---------|-------------|
| [Client](/examples/tools/mcp/sse-transport/client) | Connect an OpenAI agent to an SSE MCP server with MCPTools, and to SSE plus stdio servers together with MultiMCPTools. |
| [Client](/examples/tools/mcp/sse-transport/client) | Connect an OpenAI agent to an SSE MCP server with MCPTools, and to SSE plus stdio servers together with one MCPTools instance per server. |

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Same situation as the overview rows: the linked client page is generated (source: frontmatter) and sits on the regen queue in the PR body. Editing it by hand is overwritten at the next regeneration; the description matches the post-regen state.

| Example | Description |
|---------|-------------|
| [Client](/examples/tools/mcp/streamable-http-transport/client) | Show how to connect to MCP servers that use either SSE or Streamable HTTP transport using our MCPTools and MultiMCPTools classes. |
| [Client](/examples/tools/mcp/streamable-http-transport/client) | Show how to connect to MCP servers that use either SSE or Streamable HTTP transport using our MCPTools class. |

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Same as the SSE overview: the linked client page is generated (source: frontmatter) and on the regen queue in the PR body. The description matches the post-regen state.

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
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