OpenClaw tightens the “what actually happened?” layer for agent operations
A new May 10 OpenClaw cluster is less flashy than the beta release, but important for teams running agents through real channels. PR #80151 adds structured delivery outcomes for `openclaw agent --json --deliver`: `sent`, `suppressed`, `partial_failed`, and `failed`, including per-payload results when durable delivery can provide them. PR #80217 makes Codex-native tools visible to the diagnostic watchdog, so a long-running native bash or scraper is treated as active work rather than an abandoned embedded run. PR #80251 fixes session reset so a new session id also rotates generated transcript files and clears stale compaction checkpoints. PR #80250 adds doctor warnings when a channel-routed agent is missing the `message` tool, turning a confusing “the platform cannot do that” failure into a configuration warning.
Agent operations fail in gray zones: a message may be intentionally suppressed, half-sent, retryable, or already delivered; a tool may look idle while it is still running natively; a reset may appear clean while old compaction state remains. These PRs make those states more observable and easier to automate around.
- PR #80151 introduces an optional `deliveryStatus` object with terminal sent/suppressed outcomes and retryable partial_failed/failed outcomes for JSON delivery callers
- PR #80217 emits trusted diagnostic started/completed/error/blocked events for Codex-native tool items so watchdog logic can distinguish active native tools from stale embedded runs
- PR #80251 rotates generated session transcript paths on reset and clears compaction state that could otherwise pull an old heavy checkpoint lineage into the new session
- PR #80250 adds doctor preview warnings and prompt guidance for agents routed from channels but configured without the `message` tool
- These are active PRs at aggregation time, so operators should verify merge/release status before relying on them
- More precise delivery state helps retry logic, but only if downstream scripts stop treating all non-errors as successful sends
- Watchdog diagnostics intentionally avoid logging command arguments or output; teams still need separate process observability for deep debugging