From f2f929f59e366a69015e13bf44cb6eb5e8893547 Mon Sep 17 00:00:00 2001 From: Search Date: Mon, 6 Jul 2026 16:59:13 +0000 Subject: [PATCH] docs(claude): refresh hindsight guidance Co-Authored-By: Paperclip --- CLAUDE.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/CLAUDE.md b/CLAUDE.md index b2c2769e7e..f6e2de8743 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -72,6 +72,9 @@ cd hindsight-control-plane && npm run dev ./scripts/benchmarks/run-perf-test.sh --scale tiny # Quick smoke test ./scripts/benchmarks/run-consolidation.sh +# GitHub Actions perf artifacts intentionally retain for 1 day; durable +# history is published to the performance dashboard instead. + # Results viewer ./scripts/benchmarks/start-visualizer.sh # View results at localhost:8001 ``` @@ -121,6 +124,10 @@ PostgreSQL with pgvector. Schema managed via Alembic migrations in `hindsight-ap Key tables: `banks`, `memory_units`, `documents`, `entities`, `entity_links` +### Helm Operations + +Helm liveness probes must stay process-local so database pressure does not restart healthy pods: API liveness uses `/version`, worker liveness uses `/metrics`, and readiness remains DB-backed on `/health`. The default embedded-PostgreSQL chart values intentionally cap API DB pools and worker/retain concurrency; do not raise those defaults without validating connection pressure under retain/consolidation backlog. + ### Adding Database Migrations Hindsight runs the same Alembic tree against PostgreSQL and Oracle 23ai. Each