docs(agnote): record the MAI-UI serving lane + correct the crawl-gap read - #2475
Conversation
|
Warning Review limit reached
Next review available in: 16 minutes You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d1a441e56b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
637856f to
257308d
Compare
POWERFULMOVES
left a comment
There was a problem hiding this comment.
Field verification of #2479 — triggering a pull_request_review event now that the inline-mint fix is on main. Expect collect to get past checkout for the first time.
…read
Narrowed and corrected on Codex review before merge. Three findings, all
verified against origin/main rather than taken on the reviewer's word:
1. The first draft asserted Agent Zero's "full MCP registry" has "no fetch, no
crawl". FALSE. mcp_server.py:653 registers `e2b.surf.scrape` ("Scrape/web
surf a URL and extract content"), implemented at :567, dispatched at
:718-719. Corrected in the entry.
2. What IS broken is sharper than the original claim. `e2b_surf_scrape` POSTs to
`{E2B_MCP_SERVER_URL}/surf/scrape`, default `http://e2b-mcp-server:7073`
(mcp_server.py:61). That service is not in pmoves/docker-compose.yml, is not
running, and the `pmoves-e2b-mcp-server` fork contains no `surf` route at all
— it is a pnpm/TypeScript workspace with no endpoint of that shape. The
command is registered and dispatchable; the endpoint was never built.
3. The scope "close the research/crawl hole" over-reserved a lane this work does
not close — the three bounded items are MAI-UI serving stabilization and do
not touch `_process_crawl`. Scope narrowed to MAI-UI serving; the crawl-gap
lane is explicitly marked OPEN so it does not deter another agent.
Also records, rather than hides, that the entry was written 2026-08-07 and not
PR'd until 2026-08-08, so it gave no collision protection during the work it
describes. It is a record, not a live reservation.
Archon's half re-verified unchanged: `_process_crawl` still echoes `fragments`
and `extracted_text` from the incoming request and publishes
`status: "completed"`. A crawl request returns an empty result and reports
success.
Net effect for the next agent: evaluate completing the E2B surf backend before
building a parallel VL-fetch path — the MCP command, the dispatch, and the
fork's Dockerfile already exist.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
257308d to
6c69632
Compare
POWERFULMOVES
left a comment
There was a problem hiding this comment.
Field verification pass 3 for #2479 — this branch is now rebased onto main, so the head carries the inline-mint workflow. pull_request_review resolves the workflow from the PR head, which is why the first two attempts still ran the old file.
What
One CLAIM entry in the Active Claim Register for the crawl-gap lane. Docs only, 17 lines, no code.
This branch was pushed on 2026-08-07 and never got a PR — landing it so the lane is on the board rather than sitting unclaimed on origin. Claiming up front is the point: the register already records a prior lane that ran unclaimed and collided with Mavis.
The gap it records
Agent Zero (MC) has no fetch or crawl tool — its MCP registry is
geometry.*,ingest.youtube,media.transcribe,comfy.render,notebook.search,form.get/switch. By design it calls Archon (Master DJ) for research.Archon's half of that contract is real:
archon.crawl.request.v1/archon.crawl.result.v1are registered intopics.jsonwith closed schemas, the ARCHON JetStream stream is live (#2397, verified on 5090), and the dispatcher_handle_crawl_request→_process_crawlexists (restored in #2464, 39/39 tests green)._process_crawldoes not crawl. Re-verified againstorigin/maintoday:It reads those two fields out of the incoming request metadata and republishes them as the result with
status: "completed". There is no crawler anywhere in the repo — zerocrawl4ai/firecrawl/scrapyin any Python file; the only textual hits are an archived AGNOTE and an audit dashboard.So a crawl request returns an empty result and reports success. That is the same silent-success class this session has been correcting elsewhere.
Status since the claim was written
The entry's bounded work items (1) compose home for
mai-ui-agent/Dockerfile, (2)MAI_UI_BACKEND=local|remoteswitch, (3) CUDA guard — are all implemented in #2468, which is open and awaiting Archon review. The claim text is left at its original 2026-08-07T02:00Z wording because it is a dated register entry, not a status page; the RELEASE belongs on #2468 when it merges.Wiring
_process_crawlto anything is explicitly out of this lane — operator sequencing is stabilize before wiring.Risk
Zero. Single append to
AGNOTE4482PHI.t1.md, no RELEASE line, no other agent's lane touched.🤖 Generated with Claude Code