initTestSessionDb() creates an in-memory inbound singleton, but
openInboundDb() always opened the hardcoded /workspace/inbound.db
path. Every test that exercised getPendingMessages — directly, or via
test fixtures that load data through it (e.g. poll-loop.test.ts:29
loads formatter test rows via getPendingMessages) — failed with
SQLITE_CANTOPEN under `bun test` outside a real container.
Baseline on main: 34 pass, 25 fail across 6 files. After this fix:
59 pass, 0 fail.
In test mode, openInboundDb returns the in-memory singleton. The
singleton's .close() is no-op'd in initTestSessionDb so caller
try/finally cleanup doesn't tear down the shared DB; closeSessionDb
invokes the saved original close to do the real teardown.
Production behavior is unchanged — _inboundIsTest only flips inside
initTestSessionDb, which is never called outside the test runner.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The follow-up poller filtered /clear out of every tick without acking
the row, and pushed every other slash command through plain
formatMessages() (XML wrapping). On a warm container the outer
while(true) loop never regains control, so:
- /clear sat pending in messages_in forever (no response at all)
- /compact, /cost, /context, /files, /remote-control arrived at the
SDK as XML-wrapped user text and were never dispatched as commands
Both modes are invisible to host monitoring: rows are either left
pending without a processing_ack claim, or marked completed normally;
heartbeat keeps firing inside the SDK event loop.
When the follow-up poller observes any slash command (admin or
passthrough — categorizeMessage decides), end the active query so the
current turn winds down cleanly and the outer loop wakes, re-fetches
the same pending set, and runs them through the canonical path
(/clear handler + formatMessagesWithCommands raw dispatch). Leave the
rows untouched so the outer-loop fetch sees the same set the poller
saw.
Cost: each slash command on a warm container forces close+reopen of
the SDK stream — a few seconds of subprocess startup. The Anthropic
prompt cache is server-side with a 5-min TTL keyed on prefix hash, so
stream lifecycle does not affect cache lifetime; close+reopen within
5 min still gets cache hits.
Also corrects the warm-stream rationale comment on processQuery, which
implied keeping the stream open preserved cache warmth — it doesn't.
Testing evidence — cache stays warm across stream close+reopen:
Turn 1 (warm session):
Usage: in=6 out=245 cache_create=92 cache_read=22996
Full cache hit (22996 tokens).
Turn 2 — /clear arrives:
Pending slash command — ending stream so outer loop can process
Clearing session (resetting continuation)
Usage: in=6 out=95 cache_create=9393 cache_read=13600
System prompt + tool defs (~13600 tokens) still hit cache;
conversation history is gone (continuation reset) so the new turn
writes fresh context.
Turn 3 — /cost arrives:
Pending slash command — ending stream so outer loop can process
Usage: in=0 out=0 cache_create=0 cache_read=0 wall=0.0s api=0.0s
/cost is a CLI built-in: dispatched locally by the SDK, no API
call. Pre-fix this would have arrived as XML-wrapped user text
and never dispatched — confirms the broader fix works.
Turn 4 (next chat after /cost):
Usage: in=6 out=142 cache_create=328 cache_read=22993
Full cache hit again (22993 tokens read, 328 written). Despite the
/cost-induced stream close+reopen, the server-side prompt cache
survived: the new sdkQuery() resumed the same continuation, the
request prefix matched the cached entry.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The schedule_task MCP tool wrote routing fields (platform_id, channel_type,
thread_id) onto the outbound system message's row columns, but
handleSystemAction (src/delivery.ts) parses content JSON and forwards only
that to handlers. handleScheduleTask (src/modules/scheduling/actions.ts)
reads content.platformId/channelType/threadId — which the writer never
populated — so every kind='task' row landed in messages_in with all-null
routing.
When host-sweep wakes a scheduled task, dispatchResultText's fast path
requires routing on the message and bails when it's null, falling through
to the "Routing recovery" retry prompt. End-user delivery still works
because the agent can pick a destination from its destinations table on
retry — so the bug went undetected, silently costing one extra LLM turn
per scheduled-task wake. Sessions whose destinations table has no channel
row (e.g. agent-only destinations) fail outright with a recovery loop.
Fix: add the routing fields to the content JSON so the writer matches the
contract handleScheduleTask already expects. cancel/pause/resume/update_task
operate by id alone and don't need routing.
