Note from the human: Hi, I investigated this in depth using Codex, iterating over all possible configs to validate that this is not an isolated issue that I introduced by making adjustments to settings.

Summary

Anthropic prompt caching appears to miss on most turns in auto-reply flows.

Observed pattern across many adjacent turns:

• cacheWrite very high (~137k)
• cacheRead = 0
• small fresh input tokens

This causes repeated high-cost re-caching.

Environment

• OpenClaw: 2026.2.16 (db3480f)
• Provider: direct Anthropic (not OpenRouter)
• Model: anthropic/claude-sonnet-4-5-20250929
• Install: local CLI install
• Channel observed: Telegram auto-reply (code path appears shared across other channel adapters too)

Relevant config

• active Anthropic model has params.cacheRetention: "long"
• agents.defaults.contextPruning.mode: "cache-ttl"
• agents.defaults.contextPruning.ttl: "55m"
• agents.defaults.contextPruning.minPrunableToolChars: 50000
• agents.defaults.contextPruning.softTrim.maxChars: 4000

Expected

Within TTL, adjacent turns should show meaningful cacheRead for unchanged prompt-prefix segments.

Actual

Most turns show near-full cacheWrite and cacheRead=0.

Additional runtime signal

There are occasional cache-hit continuations immediately after tool/result boundaries, but normal subsequent user turns return to cacheRead=0 + large cacheWrite. That suggests caching is not globally disabled, but prefix reuse is unstable across regular turns.

Suspected cause

Auto-reply builds extra system prompt content from inbound metadata:

• src/auto-reply/reply/get-reply-run.ts
• src/auto-reply/reply/inbound-meta.ts

The inbound metadata includes volatile per-message fields:

• message_id
• reply_to_id
• history_count

If these are part of the cached prefix, they can change every inbound turn and defeat cache reuse.

Related issues / PRs

• Anthropic Prompt Caching Not Working - Missing cache_control Headers #17537
• [Bug]: fix(agents): add cache_control breakpoint to system prompt on OpenRouter Anthropic path #15151
• fix(cache): inject cache_control into system prompt for OpenRouter Anthropic (#15151) #17473
• fix: Anthropic Prompt Caching Not Working - Missing cache_control Headers #17560
• fix: enable prompt caching for OpenRouter passthrough with Anthropic models #11991

Request

Please confirm whether inbound trusted metadata intended for routing/reactions should remain in system-prompt cached prefix, or be segmented/moved so prompt caching remains stable across normal adjacent turns.

Happy to test candidate fixes and report before/after cacheRead/cacheWrite.