docs: refresh user-facing docs for ADR-0030 (AI gateway extensions)#83
Merged
docs: refresh user-facing docs for ADR-0030 (AI gateway extensions)#83
Conversation
Brings every operator-facing surface up to the post-merge state of the ai-proxy + cel work: - guide/dispatchers.md: full rewrite of the ai-proxy section — caller- owned model, three protocol surfaces, 4-step resolution chain, glob routes, allow/deny catalog policy with the no-fallthrough escape hatch, models_timeout_ms, Responses API specifics (preflight checks, synthetic resp_<uuid-v7>, Anthropic translation, Ollama 400), /v1/ models partial-failure shape, full metrics + error tables, six worked examples. - guide/ai-gateway.md (new): cross-cutting quickstart that ties the shipped spec fragment, the dispatcher, and the four AI middlewares together; includes the "where each concern lives" decision table. - guide/middlewares/ai-gateway.md: clarifies dispatcher vs. middleware responsibilities, adds a request.body_json + on_match.deny example, cross-links the resolution chain. - guide/middlewares/authorization.md: adds request.body_json to the cel context table; documents on_match.deny (status / code / message); adds a per-tier model-gating example. - guide/spec-configuration.md: callout for schemas/ai-gateway.yaml as a drop-in spec fragment. - guide/vacuum.md: explains why a leftover `model:` on ai-proxy now fails at lint time. - reference/extensions.md: adds the ai-proxy dispatcher reference block (caller-owned model, routes / targets / fallback, models_ timeout_ms). - ROADMAP.md: marks Responses API + /v1/models as shipped, lists the deferred items (stateful Responses, SSE re-encoding, /v1/models caching) under Someday/maybe. - README.md: rewrites the AI gateway bullet + spec snippet to use routes (no more model: in config) and references both ADR-0024 and ADR-0030. - docs/index.md, SUMMARY.md: add the new AI Gateway quickstart entry. Verified anchors via `mdbook build`; all cross-page #anchor links in the touched files resolve.
- ai-gateway.md: add layer-ordering callout (cel deny short-circuits before dispatcher) - SUMMARY.md / index.md: move AI Gateway quickstart next to Dispatchers - dispatchers.md: tier-routing example uses request.consumer instead of raw header
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
1 participant
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.
Summary
Brings every operator-facing doc surface up to the post-merge state of the ADR-0030 work (caller-owned
model, globroutes, per-targetallow/deny, stateless Responses API,/v1/modelsaggregator,celbody_json+on_match.deny, shippedschemas/ai-gateway.yamlfragment).Tier 1 (substantive content)
docs/guide/dispatchers.md— full rewrite of theai-proxysection: caller-owned model, three protocol surfaces, 4-step resolution chain, glob routes, allow/deny catalog policy with the no-fallthrough escape hatch,models_timeout_ms, Responses preflight + syntheticresp_<uuid-v7>+ Anthropic translation matrix + Ollama 400,/v1/modelspartial-failure shape, complete metrics + error tables, six worked examples.docs/guide/middlewares/ai-gateway.md— clarifies dispatcher vs. middleware concerns, adds arequest.body_json+on_match.denyexample, cross-links the resolution chain.docs/guide/middlewares/authorization.md—request.body_jsonadded to thecelcontext table;on_match.deny(status / code / message) documented; new "Match-and-deny: per-tier model gating" example.docs/reference/extensions.md— adds theai-proxydispatcher reference block (routes/targets/fallback/models_timeout_ms).ROADMAP.md— marks Responses API +/v1/modelsas shipped under the AI/LLM gateway competitive matrix; lists deferred items (stateful Responses, SSE re-encoding,/v1/modelscaching) under Someday/maybe.README.md— rewrites the AI gateway bullet + spec snippet to useroutes(no moremodel:in config) and references both ADR-0024 and ADR-0030.Tier 2 (cross-links + new surfaces)
docs/guide/ai-gateway.md— new cross-cutting quickstart that ties the shipped spec fragment, the dispatcher, and the four AI middlewares together; includes a "where each concern lives" decision table.docs/guide/spec-configuration.md— callout forschemas/ai-gateway.yamlas a drop-in spec fragment.docs/guide/vacuum.md— explains why a leftovermodel:onai-proxynow fails at lint time.docs/index.md,docs/SUMMARY.md— add the new AI Gateway quickstart entry.Test plan
mdbook buildsucceeds with no warnings#anchorlink in the touched files resolves against the generated HTML