openclaw - 💡(How to fix) Fix QMD memory search may surface generated dreaming reports as ordinary memory results [1 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#75277Fetched 2026-05-01 05:35:54
View on GitHub
Comments
1
Participants
2
Timeline
2
Reactions
2
Author
Timeline (top)
closed ×1commented ×1

I wanted to raise a question about the interaction between QMD-backed memory search and dreaming-generated report files.

This may be expected behavior, but I noticed that when QMD is used as the memory backend and default memory indexing includes memory/**/*.md, generated dreaming reports under paths such as memory/dreaming/light/YYYY-MM-DD.md can appear in ordinary memory search results alongside MEMORY.md and regular daily memory files.

The possible concern is not that dreaming creates these files — they are useful and inspectable — but that generated/staging artifacts may be retrieved with the same apparent authority as source memory files during normal recall.

Root Cause

I wanted to raise a question about the interaction between QMD-backed memory search and dreaming-generated report files.

This may be expected behavior, but I noticed that when QMD is used as the memory backend and default memory indexing includes memory/**/*.md, generated dreaming reports under paths such as memory/dreaming/light/YYYY-MM-DD.md can appear in ordinary memory search results alongside MEMORY.md and regular daily memory files.

The possible concern is not that dreaming creates these files — they are useful and inspectable — but that generated/staging artifacts may be retrieved with the same apparent authority as source memory files during normal recall.

RAW_BUFFERClick to expand / collapse

Summary

I wanted to raise a question about the interaction between QMD-backed memory search and dreaming-generated report files.

This may be expected behavior, but I noticed that when QMD is used as the memory backend and default memory indexing includes memory/**/*.md, generated dreaming reports under paths such as memory/dreaming/light/YYYY-MM-DD.md can appear in ordinary memory search results alongside MEMORY.md and regular daily memory files.

The possible concern is not that dreaming creates these files — they are useful and inspectable — but that generated/staging artifacts may be retrieved with the same apparent authority as source memory files during normal recall.

Observed shape

With dreaming enabled, memory search may include results from several kinds of files:

  • MEMORY.md — curated long-term memory
  • memory/YYYY-MM-DD.md — ordinary daily memory notes
  • memory/dreaming/<phase>/YYYY-MM-DD.md — generated dreaming summaries / candidate reports
  • session transcript collections, when enabled

The generated dreaming reports can contain candidate summaries, assistant-authored reflections, or excerpts derived from source memories. In some searches, those derived report snippets can be surfaced instead of, or ahead of, the original source memory.

Why I’m asking

I’m not sure whether QMD-backed memory search is intended to treat dreaming reports as first-class recall material, or whether those files are primarily meant as consolidation/debug artifacts.

If they are meant as ordinary searchable memory, then this may simply be a documentation/expectation question.

If they are meant more like generated consolidation artifacts, there may be a small retrieval-hygiene gap: normal memory recall can end up citing a generated candidate/reflection rather than the underlying daily note or promoted long-term memory entry.

Possible approaches

A few possible options, depending on intended behavior:

  1. Keep current indexing behavior, but document that dreaming report files are part of normal QMD recall when default memory indexing is enabled.
  2. Exclude memory/dreaming/** from default QMD memory collections unless explicitly opted in.
  3. Keep them indexed, but tag/rank them as generated/derived artifacts so ordinary recall prefers MEMORY.md and daily memory files.
  4. Add a dedicated source/filter for dreaming artifacts, so they remain searchable for debugging/review without competing with normal recall results.

I don’t have a strong preference yet; mostly opening this to confirm the intended model and see whether others using QMD-backed memory with dreaming enabled have run into the same shape.

Environment / notes

  • Memory backend: QMD
  • Dreaming enabled
  • Default memory indexing includes MEMORY.md and memory/**/*.md
  • The behavior is most visible when generated light/dreaming files accumulate enough candidate text to match ordinary recall queries

extent analysis

TL;DR

Exclude memory/dreaming/** from default QMD memory collections to prevent generated dreaming reports from appearing in ordinary memory search results.

Guidance

  • Review the current memory indexing configuration to understand how memory/dreaming/** files are being included in search results.
  • Consider excluding memory/dreaming/** from default memory indexing to maintain a clear distinction between source memory files and generated artifacts.
  • If generated dreaming reports are intended to be part of normal recall, explore options for tagging or ranking them as derived artifacts to prevent them from being prioritized over source memory files.
  • Evaluate the need for a dedicated source/filter for dreaming artifacts to allow for separate searching and debugging without impacting ordinary recall results.

Example

No specific code snippet is provided, as the issue is focused on configuration and intended behavior rather than a specific code implementation.

Notes

The optimal solution depends on the intended behavior of QMD-backed memory search with dreaming enabled. Further discussion and clarification may be necessary to determine the best approach.

Recommendation

Apply workaround: Exclude memory/dreaming/** from default QMD memory collections to prevent generated dreaming reports from appearing in ordinary memory search results, as this approach allows for a clear distinction between source memory files and generated artifacts.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix QMD memory search may surface generated dreaming reports as ordinary memory results [1 comments, 2 participants]