fix(ai-gateway): Update AI Entities docs#5698
Open
tomek-labuk wants to merge 154 commits into
Open
Conversation
to the product file
releases and set the canonical_urls accordingly. Log information depending on the status and whether the canonical_url is present or not.
when the page isn't canonical
canonicals and set `seonoindex = true`
shadowing entirely which isn't what we want. We want an existing page to take precende over the redirect, with this change an existing page matching the url is served first, if none exist the redirect takes effect.
it from including pages from an old major release
before the reference one
major_version If no major_version present, default to the latest major
inherit pages from the submodule
about major releases here, we just generate one for every version there is
Scope everything to the provided major release, if non given, default to the latest major release.
versioning pages with multiple major releases
how_to list and reference_list
includes ai-gateway and the content_type is 'plugin'
replace plugin with aigw_policy cards Comment out the how-to, we haven't migrated it yet
- metadata - links to policies instead of plugins
- remove aigw v1 how-to link - replace plugin links with policies
AIGWPolicySchema now resolves its schema file directly from app/_schemas/ai-gateway/policies/ via a FILE_INDEX constant, removing the dependency on api_plugin.data['schema']. Plugin Schema uses a lazy Hash.new FILE_INDEX that builds a per-version lowercase-keyed index on first access, replacing the eager all-versions glob and fixing the Linux case-sensitivity bug (ACL.json vs Acl.json). Removes the now-dead plugin_schemas_path config key from jekyll.yml.
tomek-labuk
removed
the
ci:manual-approve:link-validation
* update streaming to v2 * Apply suggestions from code review Co-authored-by: Lucie Milan <32450552+lmilan@users.noreply.github.com> --------- Co-authored-by: Lucie Milan <32450552+lmilan@users.noreply.github.com>
jbaross
requested changes
Jun 25, 2026
| For `http` type AI Agents, requests are proxied without A2A-specific processing. For `a2a` type AI Agents, {{site.ai_gateway}} adds protocol-aware behavior on top of plain proxying: it detects A2A requests across both JSON-RPC and REST bindings, rewrites agent-card URLs so clients discover the gateway as the canonical endpoint, and emits structured A2A telemetry to {{site.konnect_short_name}} analytics and OpenTelemetry. | ||
|
|
||
| AI Agents can be created and managed through the {{site.konnect_short_name}} UI, the {{site.ai_gateway}} API, or decK: | ||
| AI Agents can be created and managed through the {{site.konnect_short_name}} UI and the {{site.ai_gateway}} API: |
Contributor
There was a problem hiding this comment.
consider being explicit about Admin API
Contributor
Author
There was a problem hiding this comment.
Admin API and Konnect API are two separate things @jbaross
| Because the runtime executes inside {{site.ai_gateway}}, MCP endpoints are provisioned dynamically on demand. You don't host or scale them separately, and the same authentication, traffic control, and observability features available to traditional API traffic apply to MCP traffic at the same scale. | ||
|
|
||
| AI MCP Servers can be created and managed through the {{site.konnect_short_name}} UI, the {{site.ai_gateway}} API, or decK: | ||
| AI MCP Servers can be created and managed through the {{site.konnect_short_name}} UI and the {{site.ai_gateway}} API: |
Contributor
There was a problem hiding this comment.
As above, consider:
Suggested change
| AI MCP Servers can be created and managed through the {{site.konnect_short_name}} UI and the {{site.ai_gateway}} API: | |
| AI MCP Servers can be created and managed through the {{site.konnect_short_name}} UI and the {{site.ai_gateway}} admin API: |
| For the complete set of behaviors available as an AI Policy `type`, see the [AI policies hub](/ai-gateway/policies/). | ||
|
|
||
| **AI Policies are not shared.** Each AI Policy is an independent plugin instance tied to its parent entity's lifecycle. To apply identical configuration to two AI Models, create two separate AI Policies with matching `config`. This design ensures that deleting an AI Model deletes only its own AI Policies, not configurations used by other entities. | ||
| **AI Policies are not shared.** Each AI Policy is an independent configuration tied to its parent entity's lifecycle. To apply identical configuration to two AI Models, create two separate AI Policies with matching `config`. This design ensures that deleting an AI Model deletes only its own AI Policies, not configurations used by other entities. |
Contributor
There was a problem hiding this comment.
There's some redundancy here, you could merge this into the AI Policy scopes section
Contributor
There was a problem hiding this comment.
also, are race conditions possible if a policy is referenced by more than one entity?
| > | ||
| > {{site.ai_gateway}} does not define plugin configuration schemas under the AI Policy entity. | ||
| > For each plugin you intend to use as an AI Policy `type`, look up that plugin's reference page for its `config` shape. | ||
| > {{site.ai_gateway}} does not define policy configuration schemas under the AI Policy entity. |
| config: | ||
| title: AI Policy | ||
| description: An AI Gateway plugin instance scoped globally or to a specific AI entity. Policy instances are independent. | ||
| description: A reusable policy configuration scoped globally or to a specific AI entity. Policy instances are independent. |
Contributor
There was a problem hiding this comment.
If it can only be attached to exactly one entity it is not reusable
| ### AI Policy execution order | ||
|
|
||
| A Policy attached to a Model runs on the Service of the Model's derived primitives. That Policy runs at the [priority](/gateway/entities/plugin/#plugin-priority) determined by its type, which affects when it executes relative to other Policies on the request. | ||
| An AI Policy attached to a Model runs on the Service of the Model's derived primitives. That AI Policy runs at the [priority](/gateway/entities/plugin/#plugin-priority) determined by its type, which affects when it executes relative to other AI Policies on the request. |
Contributor
There was a problem hiding this comment.
Add ticket or spreadsheet entry to migrate priority info
Co-authored-by: jbaross <james.baross@konghq.com>
fabianrbz
force-pushed
the
release/ai-gateway-2.0
branch
from
060a5e2 to
7d7d76c
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Fixes #issue
Preview Links
Checklist
descriptionentry in frontmatter.