Summary

Active Memory config docs recommend cacheTtlMs values up to 300000 ms (5 min), but the zod schema rejects anything greater than 120000 ms (2 min). Following the docs literally causes the gateway to crash-loop on startup.

Reproduce

In a v2026.4.11 gateway config, add Active Memory per docs example. Docs at /app/docs/concepts/active-memory.md say:

config.cacheTtlMs — Cache reuse for repeated identical queries

No maximum documented. The paste-this-in recommended snippet in the same doc explicitly uses cacheTtlMs: 300000.

Attempt to start gateway with plugins.entries.active-memory.config.cacheTtlMs = 300000:

Config invalid
File: ~/.openclaw/openclaw.json
Problem:
  - plugins.entries.active-memory.config.cacheTtlMs: invalid config: must be <= 120000

Gateway enters crash-loop. Only lowering to <=120000 resolves it.

Expected

Either:

• Docs updated to document the actual maximum (120000 ms / 2 min), OR
• Schema relaxed to allow up to 300000 ms (matching recommended snippet)

Impact

We hit this yesterday during an Active Memory rollout. Container crash-looped 7+ times before we grepped docker logs for the actual schema error. Fix was a one-character edit to .openclaw.json, but the mismatch between docs and schema was the hard part.

Environment

• OpenClaw v2026.4.11
• plugins.entries.active-memory first-time setup
• zod schema validation at gateway boot

Proposed minimal fix

Docs update preferred (schema is sane — 2 min cache is fine for pre-reply memory recall). Just remove the 300000 from the "Paste This Into Your Agent" snippet in concepts/active-memory.md and add a sentence: "cacheTtlMs — max 120000 ms (2 min). Repeated identical queries within this window reuse the prior result."