agenr 1.8.2 → 1.9.1

package/CHANGELOG.md

@@ -2,6 +2,68 @@
 
 ## [Unreleased]
 
+## [1.9.1] - 2026-04-12
+
+Supersession stall-recovery and review-order guard patch release.
+
+### Changed
+
+- **Supersession review order is now enforced at the tool boundary.** The supersession query tool now blocks widening into lower-confidence subject sweeps until the same run has actually exhausted claim-key work, keeping the pass aligned with its intended review order.
+
+### Fixed
+
+- **Supersession passes no longer self-poison completion state by widening too early.** The surgeon now avoids the stuck `widenedBeforeClaimKeyExhausted` path that could leave a run unable to complete cleanly after later claim-key review caught up.
+- **Supersession completion feedback is more actionable.** Rejected `complete_pass` calls now tell the model how to count reviewed-but-intentionally-unlinked clusters via `entries_skipped`, reducing repeated no-progress bounded slices.
+
+### Validation
+
+Changes since last push to `origin/master`:
+
+- Fix surgeon supersession stall handling
+
+## [1.9.0] - 2026-04-12
+
+Surgeon autonomy, review-flow hardening, and CLI/runtime polish release.
+
+### Added
+
+- **Surgeon autonomy control plane v1.** Agenr now ships the first full surgeon autonomy control plane with stronger proposal-review flow, improved review ownership, and better runtime state handling across autonomous passes.
+- **Architecture review skill.** Added a dedicated architecture review skill to support repository-level structural review work alongside the surgeon and planning toolchain updates in this release window.
+
+### Changed
+
+- **Surgeon architecture and runtime boundaries are cleaner.** Surgeon logic was refactored across app, adapter, and domain seams, with generalized prompts across corpus domains, tighter runtime safeguards, simplified run flow, and more coherent completion/progress accounting.
+- **CLI and trace output are more structured.** Surgeon console output, startup summaries, progress rendering, and trace logging were tightened so autonomous runs are easier to monitor and debug.
+- **Local development scaffolding stays out of git status.** `.gitignore` now excludes the repo-local skills directory to keep local agent tooling from polluting release and day-to-day workflow status.
+
+### Fixed
+
+- **Autonomous surgeon passes recover more reliably.** Fixed proposal-loop stalling, autonomous backup handling, trace directory handling, exhausted-review completion, and repeat supersession-review suppression so long-running surgeon activity can finish cleanly.
+
+### Validation
+
+Changes since last push to `origin/master`:
+
+- Clean up surgeon CLI output with structured progress and chalk formatting
+- Simplify surgeon run flow
+- Fix surgeon trace directory handling
+- Compact surgeon trace logging
+- Implement surgeon autonomy control plane v1
+- Fix surgeon autonomous proposal loop
+- Fix surgeon autonomous backup and progress stalling
+- Stabilize surgeon retirement and proposal passes
+- Generalize surgeon prompts across corpus domains
+- Add first-principles planning skill
+- Tighten surgeon runtime safeguards
+- Update `.gitignore` to include skills directory
+- Remove first-principles planning skill and associated OpenAI agent configuration files
+- Finish supersession completion on exhausted reviews
+- Improve surgeon progress accounting and console output
+- Suppress repeat supersession review and polish surgeon CLI output
+- Add architecture review skill
+- Refactor surgeon architecture boundaries
+- Clarify surgeon pass startup summaries
+
 ## [1.8.2] - 2026-04-12
 
 Temporal parsing and ingest workflow polish patch release.

package/dist/adapters/openclaw/index.js

@@ -7,7 +7,7 @@ import {
   parseTuiSessionKey,
   readOpenClawSessionsStore,
   storeEntriesDetailed
-} from "../../chunk-PVYS6BMG.js";
+} from "../../chunk-I6A6DPNF.js";
 import {
   EMBEDDING_DIMENSIONS,
   ENTRY_TYPES,
@@ -24,7 +24,7 @@ import {
   resolveEmbeddingModel,
   runUnifiedRecall,
   validateTemporalValidityRange
-} from "../../chunk-SQARNOYD.js";
+} from "../../chunk-EMRMV2QR.js";
 import {
   resolveClaimSlotPolicy
 } from "../../chunk-GUDCFFRV.js";
@@ -1055,7 +1055,7 @@ function registerAgenrOpenClawTools(api, servicesPromise, logger) {
 var openclaw_plugin_default = {
   id: "agenr",
   name: "agenr",
-  version: "1.8.1",
+  version: "1.9.0",
   description: "agenr memory plugin for OpenClaw",
   kind: "memory",
   contracts: {

package/dist/{chunk-SQARNOYD.js → chunk-EMRMV2QR.js}

@@ -1932,6 +1932,10 @@ var CREATE_SURGEON_RUN_PROPOSALS_REVIEW_STATUS_INDEX_SQL = `
   CREATE INDEX IF NOT EXISTS idx_surgeon_run_proposals_review_status
   ON surgeon_run_proposals(review_status)
 `;
+var CREATE_SURGEON_RUN_PROPOSALS_OPEN_ISSUE_INDEX_SQL = `
+  CREATE INDEX IF NOT EXISTS idx_surgeon_run_proposals_open_issue
+  ON surgeon_run_proposals(review_status, group_id, issue_kind)
+`;
 var CREATE_META_TABLE_SQL = `
   CREATE TABLE IF NOT EXISTS _meta (
     key TEXT PRIMARY KEY,
@@ -2055,6 +2059,7 @@ var SCHEMA_STATEMENTS = [
   CREATE_SURGEON_RUN_PROPOSALS_GROUP_ID_INDEX_SQL,
   CREATE_SURGEON_RUN_PROPOSALS_CREATED_AT_INDEX_SQL,
   CREATE_SURGEON_RUN_PROPOSALS_REVIEW_STATUS_INDEX_SQL,
+  CREATE_SURGEON_RUN_PROPOSALS_OPEN_ISSUE_INDEX_SQL,
   CREATE_META_TABLE_SQL,
   CREATE_ENTRIES_CONTENT_HASH_INDEX_SQL,
   CREATE_ENTRIES_NORM_CONTENT_HASH_INDEX_SQL,

package/dist/{chunk-PVYS6BMG.js → chunk-I6A6DPNF.js}

@@ -22,7 +22,7 @@ import {
   readOptionalString,
   readRequiredString,
   validateTemporalValidityRange
-} from "./chunk-SQARNOYD.js";
+} from "./chunk-EMRMV2QR.js";
 import {
   compactClaimKey,
   describeClaimKeyNormalizationFailure,