Roadmap: 0.4.26.1 follow-ups (Tier 1 → 3 triage)

[patrick-chinchill]

Tracking issue for the 8 open gaps called out in the 0.4.26.1 release notes. None are consumer-blocking; this issue exists to make the "what ships next" decision quick in a fresh session.

How to use this issue

Pick a single item, open an implementation-planning session referencing this issue by URL, start with: "Survey chinchill-api usage for code paths that touch <item>. If none, confirm this is pure upstream-parity work and propose scope." Don't batch items — they're independent and each deserves its own PR.

Tier 1 — cheap, ship as 0.4.26.2

Target: land within a few days, one PR, short CHANGELOG.

• Port Thread.getParticipants() method #54 — Thread.getParticipants(). New public API. ~30 lines. Test via MockAdapter.
• Global handler-dispatch bound (reactions/actions/slash/modals) #61 — Global handler-dispatch bound across process_reaction / process_action / process_slash_command / process_modal_*. Extend the existing _concurrent_semaphore to cover non-message paths. Worth confirming upfront whether the right shape is "one shared bound" or "per-event-type bounds" — ask once, then implement.

Tier 2 — medium scope, own PR each

Target: one PR per item, CHANGELOG entry, no release per item (bundle 2–3 into 0.4.26.3).

• Port onOptionsLoad handler for dynamic select dropdown population #50 — onOptionsLoad handler for dynamic select dropdowns. Touches webhook dispatch in Slack/Teams + handler registration in chat.py. Pure upstream-parity.
• Port rehydrate_attachment to Adapter protocol + _rehydrate_message #52 — rehydrate_attachment adapter hook for queue/debounce + attachments. Touches Thread serialization + all 8 adapters' attachment handling. Largest test burden of this tier.
• Teams adapter: port certificate-based mTLS authentication #58 — Teams certificate-based mTLS auth. Isolated to teams/adapter.py. Requires thinking through cert storage + rotation ergonomics.
• Google Chat adapter: implement file upload support #59 — Google Chat file uploads. Upstream has a TODO here too; decide whether to lead or mirror. Isolated to google_chat/adapter.py.

Tier 3 — dedicated project

• Discord adapter: port native Gateway WebSocket support (HTTP-only today) #57 — Discord native Gateway WebSocket. New persistent-connection transport, heartbeat/reconnect logic, identify/resume flow, event dispatcher. Not a few-days PR; plan as a multi-session engagement with its own design doc.

Blocked on feature completion

• Test fidelity check should fail CI when upstream tests aren't ported #53 — 40 upstream tests without Python equivalents (Options Load, Plan variants, StreamingPlan options, getParticipants). Will start closing automatically as Tier 1/2 items land; not actionable on its own.

Prioritization signals

• None of these have reported consumer incidents. Priority comes from upstream-parity pressure (staying close to vercel/chat) and whether chinchill-api hits the code path.
• chinchill-api started depending on chat_sdk recently but its webhook surface (Slack-primary) doesn't obviously exercise any of these. Confirm before picking items — the first implementation session should grep chinchill-api for the relevant symbols.
• Divergence tax: every month on different versions makes future upstream syncs harder. Tier 1 + 2 close most of the parity delta from 4.26.

Starting a fresh session

Recommended opening prompt for the next session:

Pick up the roadmap in issue #. Survey chinchill-api under /Users/patrickliang/GitHub/chinchill-api for usage of any of the listed symbols (getParticipants, onOptionsLoad, rehydrateAttachment, Gateway, Teams mTLS, Google Chat file uploads). Report which items are on their path; then propose scope for whichever Tier 1 item we pick off first.