fix(memory): stop light dreaming from restaging stale summaries by qingminglong · Pull Request #97446 · openclaw/openclaw

qingminglong · 2026-06-28T07:37:26Z

Closes #72096

What Problem This Solves

Fixes an issue where users running memory-core dreaming would see the same light-phase work summary staged again in later cycles when the underlying daily memory content had not changed.

Why This Change Was Made

Light dreaming now filters already-emitted recall entries unless the recall freshness moved past the last light emission. Repeated same-query daily signals keep their original freshness when the stored snippet is unchanged, including cross-day re-ingestion, so unchanged daily ingestion no longer makes stale content look new while real same-day note edits can still be staged.

User Impact

DREAMS.md light-phase sections stop repeating unchanged work-summary blocks across hourly or cross-night cycles, reducing duplicate memory noise without hiding newly edited daily-note content from later light dreaming cycles.

Evidence

Claim proved: unchanged light dreaming candidates are not restaged in a later cycle, cross-day unchanged daily re-ingestion increments dailyCount without refreshing lastRecalledAt, and same-day daily-note content changes are still restaged.
Duplicate PR checks: gh search prs --repo openclaw/openclaw --state open --match title,body -- "72096", "Dreaming light-phase work summary repeats", and "light dreaming lastLightAt" returned no open PR coverage before opening this PR.

Real behavior proof: ran runDreamingSweepPhases from extensions/memory-core/src/dreaming-phases.ts against a temporary memory workspace and temporary SQLite-backed memory-core state store on PR head 5c3f7ed6f3ce3ea4566f29e7402792d6a3af729a. This exercised the real light-dreaming runtime write path into memory/2026-04-05.md with inline dreaming output; no private local OpenClaw home, credentials, or user memory were used.

Redacted terminal output from the repeated light-dreaming cycle:

{
  "proof": "PR #97446 real light-dreaming repeated-cycle proof",
  "head": "5c3f7ed6f3ce3ea4566f29e7402792d6a3af729a",
  "workspace": "<temp-workspace>",
  "stateDir": "<temp-state>",
  "cycles": [
    {
      "label": "cycle 1: initial daily note",
      "now": "2026-04-05T10:01:00.000Z",
      "candidateCount": 1,
      "noNotableUpdates": false,
      "lightBlock": "- Candidate: Added primary issue extraction for pain notifications.; Updated signals cron notification style.\n  - confidence: 0.62\n  - evidence: memory/2026-04-05.md:3-4\n  - recalls: 0\n  - status: staged",
      "logs": [
        "[info] memory-core: light dreaming staged 1 candidate(s) [workspace=<temp-workspace>]."
      ]
    },
    {
      "label": "cycle 2: unchanged note",
      "now": "2026-04-05T11:01:00.000Z",
      "candidateCount": 0,
      "noNotableUpdates": true,
      "lightBlock": "- No notable updates.",
      "logs": []
    },
    {
      "label": "cycle 3: edited note",
      "now": "2026-04-05T12:01:00.000Z",
      "candidateCount": 1,
      "noNotableUpdates": false,
      "lightBlock": "- Candidate: Added primary issue extraction for pain notifications.; Updated signals cron notification style.; Documented the shared pain notification issue.\n  - confidence: 0.62\n  - evidence: memory/2026-04-05.md:3-5\n  - recalls: 0\n  - status: staged",
      "logs": [
        "[info] memory-core: light dreaming staged 1 candidate(s) [workspace=<temp-workspace>]."
      ]
    }
  ]
}

Observed result: the first cycle staged one light candidate, the second unchanged cycle wrote - No notable updates. with candidateCount: 0, and the third cycle after editing the daily note staged the updated candidate again.

Commands / artifacts:
- git diff --check passed.
- node scripts/run-vitest.mjs extensions/memory-core/src/dreaming-phases.test.ts extensions/memory-core/src/short-term-promotion.test.ts passed: 2 files, 130 tests.
- corepack pnpm exec oxfmt --check --threads=1 extensions/memory-core/src/dreaming-phases.ts extensions/memory-core/src/dreaming-phases.test.ts extensions/memory-core/src/short-term-promotion.ts extensions/memory-core/src/short-term-promotion.test.ts passed.
- node scripts/run-tsgo.mjs -p test/tsconfig/tsconfig.extensions.test.json --incremental --tsBuildInfoFile .artifacts/tsgo-cache/extensions-test.tsbuildinfo passed, matching the previous failing CI check-test-types extension-test lane.
- python .agents\skills\autoreview\scripts\autoreview --mode local passed with autoreview clean: no accepted/actionable findings reported after the cross-day freshness fix.
Not tested / proof gaps: did not run an overnight wall-clock dreaming cycle or broad local repo checks. The added real behavior proof uses a temporary workspace and calls the shipped memory-core light-dreaming runtime entry directly to exercise the repeated-cycle write path without exposing private local memory.

clawsweeper · 2026-06-28T07:46:40Z

Codex review: needs maintainer review before merge. Reviewed June 28, 2026, 4:50 AM ET / 08:50 UTC.

Summary
The PR adds a memory-core light-dreaming freshness filter, preserves lastRecalledAt for unchanged repeated deduped recall entries, and adds regression tests for unchanged, edited, and cross-day ingestion cycles.

PR surface: Source +47, Tests +115. Total +162 across 4 files.

Reproducibility: yes. source-level: current main ranks live short-term recall entries within the lookback before using prior light-emission state, and the linked report plus PR proof show the repeated-cycle behavior. I did not run a live overnight cycle in this read-only review.

Review metrics: 1 noteworthy metric.

Persistent Memory Freshness: 1 stored timestamp write behavior changed. The PR changes when lastRecalledAt advances for repeated deduped recall entries, which affects session-state semantics before merge.

Stored data model
Persistent data-model change detected: serialized state: extensions/memory-core/src/dreaming-phases.test.ts, unknown-data-model-change: extensions/memory-core/src/dreaming-phases.test.ts, unknown-data-model-change: extensions/memory-core/src/short-term-promotion.test.ts, unknown-data-model-change: extensions/memory-core/src/short-term-promotion.ts, vector/embedding metadata: extensions/memory-core/src/dreaming-phases.ts, vector/embedding metadata: extensions/memory-core/src/short-term-promotion.test.ts, and 1 more. Confirm migration or upgrade compatibility proof before merge.

Root-cause cluster
Relationship: fixed_by_candidate
Canonical: #72096
Summary: The linked issue is the canonical report for repeated light-phase DREAMS.md work-summary blocks, and this PR is the current open candidate fix while prior same-root PR attempts are closed unmerged.

Members:

canonical: [Bug]: Dreaming light-phase work summary repeats verbatim across hourly cycles, producing apparent duplicates in DREAMS.md #72096 - The issue documents the user-visible repeated DREAMS.md light-summary behavior and is referenced with closing syntax by this PR.
superseded: fix(dreaming): filter already-emitted entries in light phase to prevent verbatim repeats (fixes #72096) #93627 - Earlier same-root candidate is closed unmerged after review found the cross-day freshness path unresolved.
superseded: fix(dreaming): filter already-emitted entries in light phase to prevent verbatim repeats (fixes #72096) #93611 - Earlier same-root candidate for the same linked issue is closed unmerged.
superseded: fix(dreaming): filter already-emitted entries in light phase to prevent verbatim repeats (fixes #72096) #93192 - Earlier same-root candidate for the same linked issue is closed unmerged and was replaced by later attempts.
superseded: fix(dreaming): filter already-emitted entries in light phase to prevent duplicate summaries (fixes #72096) #93121 - Earlier same-root candidate for this issue is closed unmerged.
1 more in the report.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Merge readiness
Overall: 🦞 diamond lobster
Proof: 🦞 diamond lobster
Patch quality: 🦞 diamond lobster
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Risk before merge

[P1] The PR changes how persisted recall freshness and prior light emissions interact; if the invariant is wrong, light dreaming could either suppress edited memory content or allow stale summaries to repeat.

Maintainer options:

Land With Current Proof (recommended)
Accept the residual session-state risk based on the direct runtime proof, focused regression coverage, clean mergeability, and green checks on the current head.
Request Longer Dreaming Proof
Ask for an overnight or cross-night repeated-cycle artifact before merge if maintainers want proof beyond the temporary-workspace runtime path.

Next step before merge

No automated repair is needed; the remaining action is maintainer merge judgment on the session-state behavior change.

Security
Cleared: No concrete security or supply-chain concern found; the diff touches memory-core runtime/tests only and does not change dependencies, workflows, permissions, or secret handling.

Review details

Best possible solution:

Land the focused memory-core repair after maintainer acceptance of the session-state risk and exact-head proof/checks remain green.

Do we have a high-confidence way to reproduce the issue?

Yes, source-level: current main ranks live short-term recall entries within the lookback before using prior light-emission state, and the linked report plus PR proof show the repeated-cycle behavior. I did not run a live overnight cycle in this read-only review.

Is this the best way to solve the issue?

Yes. The patch uses the existing phase-signal store and short-term recall freshness boundary instead of adding config or a parallel state shape, and it covers the prior cross-day freshness blocker.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 9c95abd49d45.

Label changes

Label changes:

add rating: 🦞 diamond lobster: Overall readiness is 🦞 diamond lobster; proof is 🦞 diamond lobster and patch quality is 🦞 diamond lobster.
remove rating: 🐚 platinum hermit: Current PR rating is rating: 🦞 diamond lobster, so this older rating label is no longer current.

Label justifications:

P2: This is a normal-priority memory-core bug fix with limited blast radius in DREAMS.md and future memory context.
merge-risk: 🚨 session-state: The diff changes persisted recall freshness and light-emission gating, which can stale or repeat user-visible memory context if wrong.
rating: 🦞 diamond lobster: Overall readiness is 🦞 diamond lobster; proof is 🦞 diamond lobster and patch quality is 🦞 diamond lobster.
status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (terminal): The PR body includes redacted terminal output from the real light-dreaming runtime entry showing the first cycle stages, the unchanged second cycle writes no notable updates, and an edited third cycle stages again.
proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes redacted terminal output from the real light-dreaming runtime entry showing the first cycle stages, the unchanged second cycle writes no notable updates, and an edited third cycle stages again.

Evidence reviewed

PR surface:

Source +47, Tests +115. Total +162 across 4 files.

View PR surface stats

Area	Files	Added	Removed	Net
Source	2	52	5	+47
Tests	2	116	1	+115
Docs	0	0	0	0
Config	0	0	0	0
Generated	0	0	0	0
Other	0	0	0	0
Total	4	168	6	+162

What I checked:

Repository policy read: Root AGENTS.md and extensions/AGENTS.md were read; bundled-plugin and session-state review guidance applies to this memory-core PR. (AGENTS.md:1, 9c95abd49d45)
Current main lacks light freshness gating: runLightDreaming on current main reads live short-term recall entries within lookback and ranks them before any prior light-emission filter is applied. (extensions/memory-core/src/dreaming-phases.ts:1690, 9c95abd49d45)
PR head applies phase-signal freshness before ranking: The branch wraps lookback-filtered recall entries in filterFreshLightDreamingEntries before live-file filtering and light ranking. (extensions/memory-core/src/dreaming-phases.ts:1693, 5c3f7ed6f3ce)
PR head preserves unchanged repeated freshness: The branch keeps lastRecalledAt unchanged when the same deduped query hash and snippet repeat, so unchanged cross-day ingestion can avoid looking fresh while edited content still refreshes. (extensions/memory-core/src/short-term-promotion.ts:1460, 5c3f7ed6f3ce)
Fresh light filter implementation: filterFreshLightDreamingEntries reads the existing phase-signal store and includes only entries with no prior light hit or lastRecalledAt newer than lastLightAt. (extensions/memory-core/src/short-term-promotion.ts:1786, 5c3f7ed6f3ce)
Regression coverage: PR tests cover unchanged candidates not restaging, same-day edited notes restaging, and duplicate daily signals preserving the original lastRecalledAt. (extensions/memory-core/src/dreaming-phases.test.ts:478, 5c3f7ed6f3ce)

Likely related people:

vignesh07: Introduced the dreaming sweep, diary surface, and central memory-core dreaming files that contain the changed light-dreaming path. (role: feature introducer; confidence: high; commits: 61e61ccc182f; files: extensions/memory-core/src/dreaming-phases.ts, extensions/memory-core/src/short-term-promotion.ts, extensions/memory-core/src/dreaming-narrative.ts)
SebTardif: Authored the merged REM light-staged phase-signal work using the same phase-signal store this PR now consumes for light freshness. (role: phase-signal contributor; confidence: high; commits: 8b42771aabd2; files: extensions/memory-core/src/dreaming-phases.ts, extensions/memory-core/src/short-term-promotion.ts, extensions/memory-core/src/short-term-promotion.test.ts)
obviyus: Authored the merged daily-memory re-ingestion change in the dreaming path implicated by cross-day recall freshness. (role: daily-ingestion contributor; confidence: medium; commits: 8faf91a2a8c9; files: extensions/memory-core/src/dreaming-phases.ts, extensions/memory-core/src/dreaming-phases.test.ts)
vincentkoc: Recently touched short-term promotion and light-confidence behavior in the same memory-core area, and prior review context identified the deduped daily-ingestion freshness blocker this branch addresses. (role: recent area contributor and reviewer; confidence: high; commits: 077cfca22972, 8a4a63ca076e; files: extensions/memory-core/src/short-term-promotion.ts, extensions/memory-core/src/dreaming-phases.ts)

What the crustacean ranks mean

🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

qingminglong · 2026-06-28T08:13:49Z

@clawsweeper re-review

clawsweeper · 2026-06-28T08:13:52Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

qingminglong · 2026-06-28T08:35:55Z

@clawsweeper re-review

…claw#97446) * fix(memory): stop restaging stale light dreams * fix(memory): preserve cross-day light freshness

fix(memory): stop restaging stale light dreams

bade273

openclaw-barnacle Bot added extensions: memory-core Extension: memory-core size: S labels Jun 28, 2026

fix(memory): preserve cross-day light freshness

5c3f7ed

vincentkoc merged commit 5e1f4c1 into openclaw:main Jun 28, 2026
133 of 140 checks passed

github-actions Bot mentioned this pull request Jun 28, 2026

📡 Upstream Digest — 2026-06-28 19:05 UTC curtismercier/openclaw-mods#1168

Open

SunnyShu0925 mentioned this pull request Jun 29, 2026

fix(memory-core): harden dreaming daily-file writes and drain dangling recall refs #94537

Closed

23 tasks

QiuYuang pushed a commit to QiuYuang/openclaw that referenced this pull request Jul 1, 2026

fix(memory): stop light dreaming from restaging stale summaries (open…

3e5d055

…claw#97446) * fix(memory): stop restaging stale light dreams * fix(memory): preserve cross-day light freshness

clawsweeper Bot mentioned this pull request Jul 4, 2026

fix(memory-core): prevent cross-night replication in light dreaming #100011

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

fix(memory): stop light dreaming from restaging stale summaries#97446

fix(memory): stop light dreaming from restaging stale summaries#97446
vincentkoc merged 2 commits into
openclaw:mainfrom
qingminglong:codex/72096-dreaming-light-dedupe

qingminglong commented Jun 28, 2026 •

edited

Loading

Uh oh!

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading

Uh oh!

qingminglong commented Jun 28, 2026

Uh oh!

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading

Uh oh!

qingminglong commented Jun 28, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

qingminglong commented Jun 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What Problem This Solves

Why This Change Was Made

User Impact

Evidence

Uh oh!

clawsweeper Bot commented Jun 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

qingminglong commented Jun 28, 2026

Uh oh!

clawsweeper Bot commented Jun 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

qingminglong commented Jun 28, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

qingminglong commented Jun 28, 2026 •

edited

Loading

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading