Repository navigation
fix(mcp): keep hosted scrape read-only - #454
Conversation
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
d7e0168 to
6a80d5c
Compare
There was a problem hiding this comment.
All reported issues were addressed across 3 files (changes from recent commits).
Reply with feedback, questions, or to request a fix.
Fix all with cubic | Re-trigger cubic
6a80d5c to
cbd92d9
Compare
- Serve one short instructions string per surface (full, search, keyless) from src/instructions.ts. Each fits in 1,024 characters, routes search and scrape in its first 512, and names only tools the session lists. Hosted /v2/mcp selects them per session through a new fastmcp `instructionsForSession` option (pnpm patch), so API-key and OAuth sessions no longer receive the keyless text. Locally, only stdio without a key or FIRECRAWL_API_URL gets the keyless text; the HTTP transport requires one. - Accept Alexandria provider terms in the dashboard. firecrawl_scrape still reads terms with terms/show but refuses every other terms/* capability, and terms errors link an organization admin to requiresAction.url or the data sources settings page. - Annotate hosted firecrawl_scrape with readOnlyHint: true again. Hosted scrape and search scrapeOptions load a named profile with saveChanges: false; saving browser state goes through firecrawl_interact. Local scrape keeps actions and writable profiles and stays readOnlyHint: false. - Keep firecrawl_agent at readOnlyHint: false: the research agent can click, fill forms, and navigate interactive pages. - Bump to 3.26.0.
cbd92d9 to
efb8392
Compare
|
@cubic review |
There was a problem hiding this comment.
All reported issues were addressed across 17 files
Requires human review: Auto-approval blocked because this review re-detected 1 unresolved issue already reported by Cubic.
Fix all with cubic | Re-trigger cubic
Max17190
changed the title
|
@cubic review |
There was a problem hiding this comment.
1 issue found across 16 files
Confidence score: 3/5
src/index.tsrejects the documentedfirecrawl_agentterms-approval request throughfirecrawl_scrape, preventing clients from completing the flow; update the guard to allow theterms/acceptrequest.
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="src/index.ts">
<violation number="1" location="src/index.ts:2783">
P2: This shared guard makes the documented `firecrawl_agent` terms-approval flow impossible: clients are told to call `terms/accept` through `firecrawl_scrape`, but this line rejects it. Update the agent’s terms-approval contract and instructions to use dashboard acceptance, or preserve a supported acceptance path for the agent.</violation>
</file>
Reply with feedback, questions, or to request a fix.
Fix all with cubic | Re-trigger cubic
There was a problem hiding this comment.
0 issues found across 5 files (changes from recent commits).
Confidence score: 5/5
- Automated review surfaced no issues in the provided summaries.
- No files require special attention.
Requires human review: Auto-approval blocked by 2 unresolved P0–P2 issues from previous reviews.
Re-trigger cubic
|
@cubic review |
There was a problem hiding this comment.
All reported issues were addressed across 17 files
Requires human review: Restricts hosted scrape to read-only by refusing terms writes, forcing read-only profiles, restoring readOnlyHint, and adding per-session instructions; docs and tests updated. This changes the public MCP contract and terms-acceptance workflow, so it needs human sign-off.
Fix all with cubic | Re-trigger cubic
There was a problem hiding this comment.
0 issues found across 1 file (changes from recent commits).
Confidence score: 5/5
- Automated review surfaced no issues in the provided summaries.
- No files require special attention.
Requires human review: Restores hosted scrape read-only by refusing terms writes, stripping hosted browser actions, forcing profile.saveChanges=false, and swapping per-surface instructions. Requires human sign-off because it changes terms-acceptance authorization and hosted tool behavior.
Re-trigger cubic
1 participant
Why
Provider terms acceptance currently shares
firecrawl_scrapewith page and provider data retrieval. Accepting an agreement changes organization state, so the hosted tool cannot accurately advertise itself as read-only while that operation remains reachable. Named browser profiles also save changes by default unless the request explicitly disables it.Summary
firecrawl/terms/*capability exceptterms/showbefore forwarding any call, including mixed batches. Terms-required responses preserve the requirements and retry identity, expose the read-only agreement lookup, and direct an organization admin to the existing dashboard acceptance flow. Research-agent input and output guidance uses that same dashboard flow; thread approval resumes research without accepting terms.readOnlyHint: true. Hosted scrape and search omit browser actions and force named profiles tosaveChanges: false. Profile writes remain available throughfirecrawl_interact.Tool annotations describe the available operations; client approval policies still control whether a particular call prompts the user. Hosted scrape retains provider retrieval and isolated processing of retained results without changing account, provider, or website state.
Test Plan
pnpm test: all 171 tests pass.terms/showremains usable. Research-agent metadata directs acceptance to the dashboard and preserves the approval ID, provider requirements, and thread continuation contract.profile.saveChanges: false; local search preserves the requested actions and saves with the correct non-read-only annotation.pnpm exec tsc --noEmit, ESLint on changed source files, andgit diff --checkpass.