docs: clarify semantic memory key requirements

This commit is contained in:
Peter Steinberger
2026-01-13 01:53:34 +00:00
parent 4f1c6e76fd
commit d57db17300

View File

@@ -231,13 +231,16 @@ Clawdbot also runs a **silent pre-compaction memory flush** to remind the model
to write durable notes before auto-compaction. This only runs when the workspace
is writable (read-only sandboxes skip it). See [Memory](/concepts/memory).
### Why does memory search need an OpenAI API key if I already signed in with Codex?
### Does semantic memory search require an OpenAI API key?
Vector memory search uses **embeddings**. Codex OAuth only covers
chat/completions and does **not** grant embeddings access, so the upstream
memory indexer needs a real OpenAI API key (`OPENAI_API_KEY` or
`models.providers.openai.apiKey`). If you dont want to set a key, switch to
`memorySearch.provider = "local"` or set `memorySearch.fallback = "none"`.
Only if you use **remote embeddings** (OpenAI). Codex OAuth covers
chat/completions and does **not** grant embeddings access, so **signing in with
Codex (OAuth or the Codex CLI login)** does not help for semantic memory search.
Remote memory search still needs a real OpenAI API key (`OPENAI_API_KEY` or
`models.providers.openai.apiKey`). If youd rather stay local, set
`memorySearch.provider = "local"` (and optionally `memorySearch.fallback =
"none"`). We support **remote or local embedding models** — see [Memory](/concepts/memory)
for the setup details.
## Where things live on disk