Skip to content

docs(rfc): remote crews in the chat sidebar and session ownership - #16489

Merged
chenmingwei23 merged 1 commit into
mainfrom
docs/rfc-remote-crew-sidebar
Oct 3, 2026
Merged

chenmingwei23 merged 1 commit into
mainfrom
docs/rfc-remote-crew-sidebar

Conversation

@iamwhatever

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Put one draft decision on record for how connected crews appear in the chat sidebar and which machine owns a chat on a crew, so the follow-up sidebar and ownership work has a design to trace to.

Today the sidebar mixes local and peer sessions in one list, told apart by a badge. A "chat on crew" is a local slot whose turns are relayed to the peer, so one conversation has two transcripts. That split is behind several open defects: peer approvals nobody can answer (#10618), a live peer turn shown as "Turn interrupted" after a hub restart (#14585), continue/regenerate/rewind refused with 409, the executor lost on one restore path (#10826), hub-side parent rewriting for peer workers (#14907), and older peer sessions you cannot browse (#7445).

Why it matters

Each defect is being fixed one at a time by adding another mirror of peer state onto the local slot. Without a decision on ownership, every fix grows the relay further. The sidebar shape depends on the same decision.

Not a goal

What changed (motivation → approach → change)

The two transcripts cause most of the remote-chat bugs, so the RFC compares keeping them (a) with letting the crew own its session and showing it through a window (b). Option (b) wins: every action it needs already exists on the crew's chat API, which the hub can already reach. The RFC also proposes grouping the sidebar by machine, with a connection badge per crew and dimmed cached rows when a crew is offline. It lays out five follow-up waves and five open questions, including what a crew chat shows for project and workspace. The RFC index gets one row and one preamble line for it.

Backwards compatibility

Compatible: documentation only; no runtime behaviour changes.

Removes nothing: the diff adds one document and one index row.

Tests

N/A — documentation only. python3 scripts/docs_lint.py passes locally (index reachability and citations).

Manual verification

Every cited symbol and path was checked on main at e698e8ca7a: remote_bound_refusal and its four callers, relay_remote_turn's KNOWN GAP note, _apply_recent_session (no executor restore), _clean_peer_parent, _PROXY_ALLOWED_PREFIXES, the approve/continue/regenerate/rewind routes under api/chat, and TunnelState.

Related Issues

Refs #6180, #7445, #10618, #10826, #14585, #14907

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

@iamwhatever
iamwhatever requested a review from a team as a code owner October 3, 2026 10:51
@iamwhatever
iamwhatever requested a review from kyleseaman October 3, 2026 10:51
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: checking Automated validation is still running labels Oct 3, 2026
@github-actions

github-actions Bot commented Oct 3, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of f52f672e8036d038584f132eb0fb4d4c94ebfb3c — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Wave 1 closes #10618 and #14585 as superseded, so relayed peer approvals stay unanswerable until wave 3 ships. A maintainer accepting the RFC should confirm that gap is acceptable.

[DESIGN-REVIEWED] f52f672

Description read

[DESCRIPTION-READ] f2f4b8e9a6112ab98bb3152caa6616ace0bb3151ef39454d5cc4edce0977efb0

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions github-actions Bot added the readiness: action required A blocking check or review needs attention label Oct 3, 2026
@github-actions

github-actions Bot commented Oct 3, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed f52f672e8036d038584f132eb0fb4d4c94ebfb3c — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] f52f672

Verdict parsed from the review's SHA-scoped output markers for commit f52f672e8036d038584f132eb0fb4d4c94ebfb3c.

False positive or not applicable? A repository writer can comment:
/ai-review override fable f52f672e8036d038584f132eb0fb4d4c94ebfb3c: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 3, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of f52f672e8036d038584f132eb0fb4d4c94ebfb3c and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] f52f672

False positive or not applicable? A repository writer can comment:
/ai-review override gpt f52f672e8036d038584f132eb0fb4d4c94ebfb3c: <one-sentence reason>

…session ownership

Proposes per-machine sidebar groups (Local + one collapsible group per connected crew with a connection badge, cached-and-dimmed when disconnected) and peer-owned crew sessions viewed through the existing api/chat + api/stream proxy, retiring the relay-into-local-slot model in waves. Adds the index row.
@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • The redaction premise is false span=ade046fb55a9

fixed: Fixed in f52f672: the trust-boundary row now says api_instances_proxy does not redact today, and Security considerations requires wave 3 to apply redact_peer_text on the read path.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • Today's relay redacts peer text span=2b9e07bee51a

fixed: Fixed in f52f672: the (a) column now states the relay redacts with redact_peer_text before storing; (b) must keep that on its read path.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • Wave 3 reads api/chat and api/stream through the proxy span=3f5571bc924f

fixed: Fixed in f52f672: Security considerations now names the proxy as the main read path under (b) and lists what that changes.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • Peer text would reach the hub browser unredacted span=dd2733863ee3

fixed: Fixed in f52f672: wave 3's exit criteria now include that a credential planted in peer text renders redacted in the window.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • api/stream widens volume once it is the main read path span=57ae109d729f

fixed: Fixed in f52f672: Security considerations adds a Volume bullet: the window subscribes to api/stream only while a crew session is open.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • Comparison credits (b) with a control that doesn't exist span=26854a44c3b2

fixed: Fixed in f52f672: the table no longer claims existing redaction for (b); it names the redaction wave 3 must add. Direction (b) unchanged.

@iamwhatever

Copy link
Copy Markdown
Contributor Author
  • Offline cache lives in browser memory span=763dd87aa0b9

fixed: Fixed in f52f672: the Works offline row and design section 1 now say the cache is browser memory and a reload leaves a disconnected crew's group empty until it reconnects.

@iamwhatever
iamwhatever force-pushed the docs/rfc-remote-crew-sidebar branch from 8626cf6 to f52f672 Compare October 3, 2026 11:00
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 3, 2026
@chenmingwei23
chenmingwei23 merged commit 04603f3 into main Oct 3, 2026
98 checks passed
@chenmingwei23
chenmingwei23 deleted the docs/rfc-remote-crew-sidebar branch October 3, 2026 17:08
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 3, 2026
iamwhatever pushed a commit that referenced this pull request Oct 4, 2026
Record maintainer iamwhatever's acceptance of the remote-crew sidebar
RFC (option (b): the peer owns a crew session and the local dashboard
is only a window, plus per-machine sidebar groups). Status moves from
draft to accepted, doc-pr names #16489, the prose status line and the
directory README row say so, and the decision is written under the
document's Open questions with its date. Nothing is implemented on main.
chenmingwei23 pushed a commit that referenced this pull request Oct 4, 2026
Record maintainer iamwhatever's acceptance of the remote-crew sidebar
RFC (option (b): the peer owns a crew session and the local dashboard
is only a window, plus per-machine sidebar groups). Status moves from
draft to accepted, doc-pr names #16489, the prose status line and the
directory README row say so, and the decision is written under the
document's Open questions with its date. Nothing is implemented on main.

Co-authored-by: Joe Guo <zejiangg@amazon.com>
@bolichen97 bolichen97 mentioned this pull request Oct 6, 2026
5 tasks done
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants