Resources - ish

Resources are read via the MCP read_resource call. Some clients (ChatGPT) do not read resources; the docs_get tool mirrors the docs resources for them.

URI	Returns	Description
`ish://docs`	application/json	Index of every Ish concept, reference, and guide doc. Returns JSON [{slug, title, description}, …]; fetch any page by reading ish://docs/<slug>.
`ish://identity/me`	application/json	Currently-authenticated user, decoded from the validated JWT. Returns JSON {user_id, email, role}. role is either ‘admin’ or ‘user’. Use this once at session start to confirm whose workspaces the agent is acting on.
`ish://docs/{slug*}`	text/markdown	Markdown body of one Ish doc page. Slugs use forward-slashes (e.g. ‘overview’, ‘concepts/study’, ‘reference/aliases’). List every available slug via the ish://docs index resource.
`ish://reference/content-types/{modality}`	application/json	Legal content_type values for a study modality. Returns JSON {modality, values: […]}. Read before study_create when the modality is video / image / document / audio / text; interactive and chat return an empty list.
`ish://study/{study_id}/screenshots`	application/json	JSON index of every screenshot on a study, grouped by frame. Each screenshot entry carries a `resource_uri` (`ish://study/<sid>/screenshot/<scid>`) - read that to fetch the image bytes. Use this before pulling individual screenshots so the agent can pick representative frames rather than reading every one.
`ish://study/{study_id}/transcripts`	application/json	JSON list of flattened chat transcripts for a chat study - one per completed participant on external-chatbot iterations, or one per conversation on participant-pair iterations. Same shape as `study_get(view='transcripts')`. Errors on non-chat studies (no transcripts to surface).
`ish://study/{study_id}/screenshot/{screenshot_id}`	image/png	Image bytes for one screenshot on a study. The server resolves the screenshot’s storage URL (Supabase Storage) and streams the body back as a base64-encoded BlobResourceContents with the upstream mime type. Discover screenshot ids via the ish://study/<study_id>/screenshots index.
`ish://iteration/{iteration_id}/media`	application/json	JSON index of the media URLs on an iteration: content_url (video/audio/document), image_urls (image), and featured_image_url (text). Each item carries a `resource_uri` (`ish://iteration/<id>/media/<kind>/<index>`) the agent can read to fetch the bytes.
`ish://iteration/{iteration_id}/media/{kind}/{index}`	application/octet-stream	Bytes for one media item on an iteration. `kind` is one of `content` (video/audio/document content_url, index always 0), `image` (one entry of image_urls, indexed from 0), or `featured_image` (text iteration’s featured_image_url, index always 0). Discover the available items via the ish://iteration/<id>/media index.
`ish://workspace/{workspace_id}/secrets`	application/json	Metadata for every secret registered on a workspace (keys, scope, variable_type, timestamps). Values are never returned. Read this before pointing a chatbot endpoint header at {{secret:KEY}} to confirm the key exists. Secret writes happen via the ish CLI (`ish secret set`).