Health checks reuse a two-second snapshot and bound the schema query. Task SSE writes an error frame when an event cannot be encoded. ResolveDocTask errors when the frozen plan has no duty row, while explicit none-duties stay zero. Drop the unwired semantic-summary replace env, record the intentional dev versus release model split, document the test/main cutover path, and archive pre-goose SQL.
Co-authored-by: Cursor <cursoragent@cursor.com>
Also declare SEMANTIC_SUMMARY_SHADOW and PI_CHANGE_REPORT_NARRATIVE on
test/main, remove dead CURSOR_RUNNER_URL/cursor-agent, and set main
SEED_DEMO_USERS=false. Defaults stay legacy/false (no behavior change).
Co-authored-by: Cursor <cursoragent@cursor.com>
Rules that already split skip the LLM. Independent flags stay off by
default and do not raise MULTI_INTENT_PLAN_TIER.
Co-authored-by: Cursor <cursoragent@cursor.com>
Flag-on CreateAgentCandidate asks the product PI runner for
change_report_narrative_v1 after the real digest is known; timeout
and malformed payloads stay full fallback. Default flag remains off.
Co-authored-by: Cursor <cursoragent@cursor.com>
Enable SEMANTIC_SUMMARY_SHADOW to call the product PI runner on authorized messages only, keep REPLACE unwired, and record MiniMax evidence for the D3 matrix close.
Co-authored-by: Cursor <cursoragent@cursor.com>
Remote /v1/run now emits additive result_v1; Go validates run_id, digest, and paths before materializing, and only accepts missing typed payloads with an explicit legacy handshake.
Co-authored-by: Cursor <cursoragent@cursor.com>
HTTP Runtime does not start the worker. Each compose environment gets its own events service, health port, and bounded pool.
Co-authored-by: Cursor <cursoragent@cursor.com>
Wire RUNNER_AUTH_TOKEN through dev/test/main Compose, validate uploads
with extension whitelist and content sniffing, add readiness soak script
and short PG loop test, and require runner token in prod when URL is set.
Co-authored-by: Cursor <cursoragent@cursor.com>
