The signed-in user, org, plan, API URL and currently selected project. Call this first to confirm which environment the session is pointed at.

All projects in the organisation, with ids.

OpenAPI specs uploaded to a project, newest first.

Upload OpenAPI spec text (JSON or YAML) as the project's new current spec. Returns the created spec id.

Import an OpenAPI spec from a public URL.

The SDK generator's language catalogue and the preference keys each language accepts.

Every operation the parser found in the current spec, with method, path, operationId and deprecation, the list the SDK Studio renders.

The project's persisted SDK Studio settings, per-endpoint overrides, revision, and last published release. The revision is required to write settings back.

Write SDK Studio settings. Pass a full settings object (read it first with octri_get_sdk_settings and modify), plus the revision you read. A stale revision is rejected.

Run the generator's validator against the project's current spec. Cheap; do this before spending a build.

Generate one language WITHOUT spending a build, and return the emitted file paths (and optionally contents). This is the fast loop for generator changes.

Queue an SDK build for one or more languages. Returns a build id immediately; poll with octri_get_build or block with octri_wait_for_build.

Recent SDK builds with per-language status.

One build's current state, including each language's lifecycle phase and any failure message.

Block until every language of a build reaches a terminal state, then return the final per-language result. Use after octri_trigger_build.

Re-run only the named languages of an existing build, in place. Omit languages to retry exactly the ones that failed.

Download and extract a build's artifacts into the local cache so their files can be listed and read. Returns per-language file counts and a content fingerprint.

List the files of a previously fetched build+language, with sizes. Call octri_fetch_artifacts first.

Read one generated file from a fetched build. This is how to inspect what the generator actually emitted.

Compare two fetched builds file-by-file per language: which emitted files were added, removed or changed. The primary way to judge a generator change.

Documentation pages generated for the project.

One documentation page by slug.

The MCP tool catalogue this project publishes to its own customers, derived from its SDK config.

Per-language GitHub repositories linked to the project's SDKs.

Production health for the project: event count, error count, error rate and distinct issues over a window. Call this before reading issues. It tells you whether there is anything to look at.

Grouped production errors, newest first. Each issue has an id usable with octri_get_issue.

One issue with its stack frames and recent events. A frame with resolved=true was de-minified against an uploaded source map; if none are, the symbols for that release were never uploaded.

Triage an issue: resolve it, ignore it, or reopen it (status unresolved).

Raw production events, unaggregated. Use when an issue's grouped view is not enough, e.g. to see every occurrence across releases.

Error rate, new issues and regressions per release. This is how you tell whether a deploy made things worse.

Slowest transactions (p50/p95) and suspected N+1 query patterns for the project.