fix(v2): a2a-DM notice render glitch + get_context recent messages (reactions) - #626
Merged
Merged
Conversation
Two things surfaced looking at a pod with an agent-to-agent DM: 1. The "X and Y started a DM" system notice rendered as a squished 38px sliver — no headline, wrapped timestamp, bare "Open conversation" button. Cause: .v2-msg is a 2-col grid (38px avatar | 1fr body); the notice has a single child (.v2-syscard), so it landed in the avatar column. Fix: .v2-msg--system is now display:block so the card fills the row. Guarded in the v2 layout invariants test. Verified in-browser (rule 9). 2. Local agents rarely react to messages. The capability works end-to-end (dual-auth reaction endpoint + commonly_react_to_message, verified live), but get_context returned NO messages despite its "recent messages" contract — so an agent orienting had no messageId to react to (or reply to a specific message). Added `recentMessages` (id, author, content, createdAt) to get_context, honoring the contract. Skill now teaches the react flow (get a messageId, then react) and to reach for reactions often.
lilyshen0722
added a commit
that referenced
this pull request
Jul 5, 2026
…rce order) (#627) The first fix (#626) set display:block on .v2-msg--system, but .v2-msg sets display:grid LATER in the file at equal specificity — so grid won on source order and the notice stayed squished into the 38px avatar column (regressed on refresh). Bump to the compound selector .v2-msg.v2-msg--system so specificity wins regardless of order. Verified in-browser WITHOUT !important this time (headline 0->481px, display grid->block). Test now guards the compound selector. Also re-applies the CONNECTING_LOCAL_AGENTS 'what a local agent can do' doc update that was orphaned from #626 (pushed after its auto-merge).
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
1 participant
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.
Two issues found looking at a pod with an agent-to-agent DM:
1. The "X and Y started a DM" notice rendered broken — a 38px sliver with no headline, a wrapped timestamp, and a bare "Open conversation" button.
.v2-msgis a 2-col grid (38px avatar | 1fr body); the system notice has a single child (.v2-syscard), so it got crammed into the avatar column. Fix:.v2-msg--system { display: block }so the card fills the row. Added a layout-invariant guard; verified in-browser (before/after screenshots).2. Local agents rarely react. The capability works end-to-end (dual-auth reaction endpoint +
commonly_react_to_message— verified live: 👍 landed withmine:true). Butget_contextreturned no messages despite its "recent messages" contract, so an agent orienting had nomessageIdto react to (or to reply to a specific message). AddedrecentMessages(id/author/content/createdAt) toget_context; the skill now teaches the react flow and to reach for reactions often.🤖 Generated with Claude Code
https://claude.ai/code/session_01MnRCAFgjrrGZxo9VRCmCm9