Feat: Storage as resources#1003
Draft
RobertCrupa wants to merge 4 commits into
Draft
Conversation
Add an MCP resources surface for Apify storage data reads alongside the existing storage tools (additive; no tool changed). Three `apify://` templates (dataset items, KVS key listing, KVS record) plus a bounded recent-list of the user's datasets/stores. Resource handlers resolve the per-request token and build an ApifyClient like the CallTool handler; no-token sessions skip storage and read returns an explanatory text block. Closes #1002 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
github-actions
Bot
added
t-ai
Contributor
There was a problem hiding this comment.
⚠️ Not ready to approve
There are correctness/contract issues in the new resources layer (unsafe typing for blob contents, URI decode error handling, and MIME/doc mismatches) that should be fixed before approval.
Pull request overview
Implements Issue #1002 by adding an MCP resources surface for Apify storage reads (datasets, KVS keys, KVS records) alongside existing storage tools, wiring the per-request token into the resources handlers and extending unit/integration coverage.
Changes:
- Add
apify://…storage resource templates, recent-resource listing, and resource reads for dataset items / KVS keys / KVS records. - Thread an optional
ApifyClientthroughresource_serviceand build it per request inActorsMcpServer. - Add unit + integration tests for the new resources surface, and cover empty KVS record bodies in the existing tool.
File summaries
| File | Description |
|---|---|
| tests/unit/tools.get_key_value_store_record.test.ts | Adds unit coverage for empty KVS record bodies to keep structured output schema-conforming. |
| tests/unit/resources.storage.test.ts | New unit tests for storage resource templates, listing, and read behavior across datasets/KVS. |
| tests/unit/resources.service.test.ts | Updates resource service template listing expectations to include storage templates. |
| tests/integration/suite.ts | Adds integration coverage for resources/templates/list, resources/list, and resources/read for storage. |
| src/tools/common/get_key_value_store_record.ts | Ensures schema-required value is present even when apify-client returns undefined for empty bodies. |
| src/resources/storage_resources.ts | Introduces apify:// storage resource parsing, recent listing, and read implementations. |
| src/resources/resource_service.ts | Extends resource service to delegate apify:// URIs to storage resources and to accept an optional ApifyClient. |
| src/resources/AGENTS.md | Documents the new storage resources surface and how it’s wired. |
| src/mcp/server.ts | Wires ListResources/ReadResource to build an ApifyClient from per-request token metadata. |
Copilot's findings
- Files reviewed: 9/9 changed files
- Comments generated: 4
Note
Your feedback helps us improve the quality of this feature.
Please use 👍 or 👎 to tell us whether this assessment is correct.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
23
to
30
| type ExtendedResourceContents = TextResourceContents & { | ||
| html?: string; | ||
| _meta?: AvailableWidget['meta']; | ||
| }; | ||
|
|
||
| type ExtendedReadResourceResult = Omit<ReadResourceResult, 'contents'> & { | ||
| contents: ExtendedResourceContents[]; | ||
| }; |
Comment on lines
+135
to
+156
| // apify://datasets/{datasetId}/items | ||
| if (segments[0] === 'datasets' && segments[2] === 'items' && segments.length === 3) { | ||
| return readDatasetItems(uri, apifyClient, decodeURIComponent(segments[1]), query); | ||
| } | ||
|
|
||
| // apify://key-value-stores/{keyValueStoreId}/keys | ||
| if (segments[0] === 'key-value-stores' && segments[2] === 'keys' && segments.length === 3) { | ||
| return readKeyValueStoreKeys(uri, apifyClient, decodeURIComponent(segments[1]), query); | ||
| } | ||
|
|
||
| // apify://key-value-stores/{keyValueStoreId}/records/{recordKey} | ||
| if (segments[0] === 'key-value-stores' && segments[2] === 'records' && segments.length === 4) { | ||
| return readKeyValueStoreRecord( | ||
| uri, | ||
| apifyClient, | ||
| decodeURIComponent(segments[1]), | ||
| decodeURIComponent(segments[3]), | ||
| ); | ||
| } | ||
|
|
||
| return buildTextResult(uri, `Resource ${uri} is not a recognized Apify storage URI.`); | ||
| } |
Comment on lines
+262
to
+265
| const { value, contentType } = record; | ||
| // apify-client maps an empty record body to `undefined`; emit empty text (an empty OUTPUT is legitimate). | ||
| if (value === undefined) return buildTextResult(uri, ''); | ||
| if (Buffer.isBuffer(value)) { |
Comment on lines
+26
to
+31
| `resources/list` adds concrete URIs for the user's recent datasets/stores | ||
| (`desc: true`, bounded). Contents are `application/json` for items/keys; records keep | ||
| their `contentType` (binary → base64 `blob`). Best-effort: no token / API error → | ||
| list omits storage; an unreadable read returns an explanatory `text` block, never an | ||
| error. Reuses the storage tools' arg-parsing helpers and 404→soft-fail pattern; it | ||
| does **not** share their response builders (resources need `ReadResourceResult`). |
Contributor
Author
|
Currently, large binaries are inlined in the resource interface. @jirispilka, we will need to decide whether to mirror the same approach with the tools, serving the URI when binaries are > 256Kb. I am in favor of letting the LLM process the data from a resource link using a script, rather than dumping it into the context. |
resources are now pulled from the api.apify.com, and large binaries return signed resource links
jancurn
reviewed
Jun 18, 2026
| - **Key-value store:** Flexible storage for unstructured data or auxiliary files. | ||
|
|
||
| ## Apify API resources | ||
| - Any Apify API GET endpoint can be read as an MCP resource. Pass the full \`https://api.apify.com/v2/...\` URL to resources/read; the server injects authentication and returns the response body. |
Member
There was a problem hiding this comment.
Suggested change
| - Any Apify API GET endpoint can be read as an MCP resource. Pass the full \`https://api.apify.com/v2/...\` URL to resources/read; the server injects authentication and returns the response body. | |
| - Any Apify API GET endpoint can be read as an MCP resource. Pass the full \`https://api.apify.com/v2/...\` URL to `resources/read`; the server injects authentication and returns the response body. |
jancurn
reviewed
Jun 18, 2026
| }); | ||
| } | ||
| } catch { | ||
| // Ignore: best-effort listing. |
Member
There was a problem hiding this comment.
Perhaps at least log the error?
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.
WIP for adding storage as a resource interface