spec(automation): kanban-card-chat -- a card opens its agent conversation - #5389
Merged
Merged
Conversation
…ns its agent conversation One conversation per (person, task), keyed task:gHashTag/<repo>#<n> in the host's agent_messages; the same thread for the app's sheet, tri card and any A2A agent (metadata.task or contextId = the key). The frame opens a card only through the named same-origin envelope; the sheet opens at half with the composer pinned; the agent's note about the card is bounded and fenced. test-report 10/10; negative control (assert(RUN_LIVE)) FAILs as it must. String constants are pinned by the host test: the Zig backend does not compare []const u8 with ==. Refs #5388 (EPIC) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
This was referenced Oct 1, 2026
This was referenced Oct 1, 2026
Merged
Merged
…, actions on a tap The sheet says which pull requests name the issue and the worst state of their checks; a read that did not happen is "not measured", never "no PR". The agent's proposals are approve/decline cards in the card chat and run only on the person's tap. The column is the Queen's verdict: no move is offered, and the agent does not open PRs (the host holds no GitHub token). test-report 14/14; negative controls on checks_rank and action_runs FAIL. Refs #5388 (EPIC) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-10-01 13:14:21 UTC
Summary
Seal Status
|
… ends, four reads per open Refs #5388
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
This was referenced Oct 1, 2026
…e; the move door's contract Refs #5388
Contributor
PR DashboardGenerated at: 2026-10-01 14:42:03 UTC
Summary
Seal Status
|
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
This was referenced Oct 2, 2026
Merged
Merged
Merged
gHashTag
added
the
bee-reviewed
Owner
Author
|
Reviewer bee W -- verified at head Evidence (fresh clone under /tmp, master 6111696, t27c built from master):
Red checks are noise: Corpus ratchet, spec-guards, untrusted-input, coverage and Scorecard are red on master too; emit-bitexact fails on every open PR and does not run on master pushes. |
This was referenced Oct 2, 2026
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.
Refs #5388 (EPIC: a card on the game's kanban opens its agent conversation).
What
specs/automation/kanban-card-chat.t27: the spec of record for "tap any card on app.t27.ai/game/kanban, a chat with the agent opens with its history, and any A2A agent continues the same thread".It pins:
task:gHashTag/<repo>#<n>, owned by this repo's owner only. A bare number takes the board's repository and is never guessed.?c=conversation may not.{v:1, type:'t27-app', kind:'open-card'}, accepted from the same origin only.NOTE_MAX1200) and fenced as foreign text. It says so when the board is down or the card is untitled.?card=<n>[&repo=].metadata.taskor acontextIdequal to the thread key continues the thread. History is capped at 20 turns.RUN_LIVE=false: the chat is designed but not yet live.Checks
t27c typecheck: OK.t27c test-report: 19/19 at v4 (16 at v3, 10 at v1).TIMELINE_PAGES_MAX), the unread middle is a floor (more_is_lower_bound), checks take one read (CHECKS_ONE_READ), and an open costs at most 4 of the 60 public reads an hour (READS_PER_OPEN_MAX). MutatingTIMELINE_PAGES_MAXto 1 fails both new tests.STATUS_DOORwithSERVER_TOKEN_ENV: the server reads once for everyone (SERVER_READS_PER_OPEN1, newestSERVER_REFS_LASTreferences, a shared cache), and with no token the client reads as in v3 (status_source).MOVE_*: a move is one card through a Queen door that does not exist yet (MOVE_DOOR_EXISTSfalse,REGISTRY_PUT_IS_A_MOVEfalse), says the column the person saw, and targets only backlog or dropped. Mutatingmove_target_okto accept done fails the move test.assert(RUN_LIVE)fails 1 test.==on[]const u8, so the string values are pinned by the host test (999render/src/spec/kanban-card-chat.test.ts).The host implementation (render task thread, A2A continuation, the app's card sheet) lands in gHashTag/999-multibots-telegraf.
🤖 Generated with Claude Code