v0.9.1-rc.2 — increased dial timeout

Increase dial timeout from 25ms to 500ms to eliminate false probe timeouts on slower systems. The original timeout was too aggressive and caused false alarms even when the daemon was running.

v0.9.1-rc.1 — socket emission fixes (testing)

Pre-release candidate for testing issue #148 fixes on EC2 before cutting 0.9.1. Install with: openclaw plugins install @agnt-rcpt/openclaw@rc --force

v0.9.1 — socket emission and startup probe fixes

Fixes issue #148 with socket emission failures and misleading startup warnings.

Changes

• Increased dial timeout from 25ms to 500ms to eliminate false probe timeouts on slower systems
• Wired debugLog to api.logger.info so per-call drops are visible in logs
• Fixed silent promise rejection swallowing in hook catch handlers
• Startup probe now only shows install hint for real errors (ENOENT, ECONNREFUSED), not timeout races

What's fixed

• Socket emission no longer silently fails on slow systems
• Transport drops are now visible in openclaw logs
• Startup warnings are no longer false alarms
• Promise rejections are logged, not swallowed

v0.9.0

What's Changed

• chore: relicense from MIT to Apache-2.0 by @ojongerius in #138
• chore(deps): bump the node-deps group across 1 directory with 5 updates by @dependabot[bot] in #137
• feat(emitter): track and flush drop_count to daemon (issue #139 stage 1) by @ojongerius in #140
• fix(emitter): defer logDrop(write) until retry also fails by @ojongerius in #142
• feat(plugin): ADR-0010 Stage 2 — thin emitter cutover by @ojongerius in #143
• docs: README and CHANGELOG for ADR-0010 Flavor B cutover (Stage 3) by @ojongerius in #144
• docs: ship CHANGELOG.md in npm tarball and document contributor workflow by @ojongerius in #145
• chore(deps): bump @agnt-rcpt/sdk-ts to 0.9.0 by @ojongerius in #146
• chore(release): bump to 0.9.0 by @ojongerius in #147

Full Changelog: v0.8.0...v0.9.0

v0.8.0

See CHANGELOG for details.

v0.8.0-alpha.1

What's Changed

• chore(release): v0.6.0 by @ojongerius in #124
• feat(emitter): add fire-and-forget AF_UNIX socket emitter (ADR-0010) by @ojongerius in #125
• ci: route pre-release tags to non-default npm dist-tag by @ojongerius in #126
• chore(release): bump version to 0.8.0-alpha.1 by @ojongerius in #127

Full Changelog: v0.6.0...v0.8.0-alpha.1

v0.6.0

What's Changed

• chore(release): v0.5.0 by @ojongerius in #117
• fix(tools): add timestamp/chain filters and newest-first ordering to ar_query_receipts by @ojongerius in #119
• chore(deps-dev): bump openclaw from 2026.4.24 to 2026.4.29 in the node-deps group by @dependabot[bot] in #120
• chore: prep v0.6.0 changelog and gitignore Claude Code state by @ojongerius in #123

Full Changelog: v0.5.0...v0.6.0

v0.5.0

What's Changed

• docs: document parameterPreview, fix profile:full example, update ecosystem link by @ojongerius in #109
• docs: improve CLI reference for subcommands, receipt querying, and parameterPreview by @ojongerius in #113
• docs: document correct gateway restart command and hot-reload behaviour by @ojongerius in #114
• chore: add release automation script by @ojongerius in #115
• chore: bump sdk-ts to 0.6.0; rename preview → disclosure by @ojongerius in #116

Full Changelog: v0.4.2...v0.5.0

v0.4.2

Fixed

• Classify sessions_spawn and subagents as system.command.execute (high risk) instead of system.application.launch (low risk). Spawning a new agent session is a high-privilege operation; receipts now reflect that in audit trails (#106).
• Scope session_start pending-stash eviction to the current session only. Previously pending.clear() wiped stashed call data for every in-flight tool call across all sessions, causing concurrent sessions to lose their startedAt timestamp and fall back to recomputing paramsHash from the event params instead of the original stash (#107).
• Scope pending eviction by (sessionKey, sessionId) pair, not sessionKey alone. Two sessions sharing a sessionKey but with different sessionIds are distinct; evicting by sessionKey alone still allowed one session's session_start to clear another's pending stash (#107).

v0.4.1

Fixed

• Recover chain state from the store after a plugin restart. When the process restarted mid-session, the in-memory sequence counter was re-initialised to 0 while the database still held prior receipts, causing every subsequent receipt insert to fail with UNIQUE constraint failed: receipts.chain_id, receipts.sequence and leaving the chain permanently stuck for that session (#103).