Propose a patch

Evidence

The sandbox observed nothing beyond the static findings; the proposal cites those.

diff --git a/examples/review-sample.md b/examples/review-sample.md
--- a/examples/review-sample.md
+++ b/examples/review-sample.md
@@ -19,1 +19,0 @@
-- **Update MEMORY.md:** Add note that `scripts/backup-remote.sh` was rewritten 2026-03-15 to use native `openclaw backup create`. Old split-workspace + separate-config approach replaced.
@@ -117,1 +116,0 @@
-- MEMORY.md doesn't mention it — no update needed
diff --git a/prompts/nightly-review.md b/prompts/nightly-review.md
--- a/prompts/nightly-review.md
+++ b/prompts/nightly-review.md
@@ -77,1 +77,0 @@
-1. Edit the relevant `memory/sections/*.md` file(s) directly when sectioned memory exists; otherwise edit `MEMORY.md`
@@ -79,1 +78,0 @@
-3. Run `cd "${CLAWD_DIR}" && git add MEMORY.md memory/index.md memory/sections && git commit -m "dreamer: auto-apply"`
@@ -101,1 +99,0 @@
-b. Git commit: `cd "${CLAWD_DIR}" && git add MEMORY.md memory/sections && git commit -m "dreamer: cleanup"`
diff --git a/scripts/trend_detection.py b/scripts/trend_detection.py
--- a/scripts/trend_detection.py
+++ b/scripts/trend_detection.py
@@ -395,1 +395,0 @@
-lines.append("_Projects in MEMORY.md not mentioned in any daily note for 30+ days._\n")

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.8.0. If the behavior is intended, declare it (host under permissions.network, a line in the README) instead of removing it; the next crawl scores declarations.

Issue / PR body

Title: lucid-dreamer: Instructs the agent to modify its persistent memory or instruction files (KillerSkills verdict caution 71)

## What the honeypot observed
KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.8.1) and scored it **CAUTION 71/100**. Summary: static analysis only — behavior not yet observed.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/clawhub/lucid-dreamer/0.8.0

## Findings with a location
- `examples/review-sample.md:117` · Instructs the agent to modify its persistent memory or instruction files — examples/review-sample.md:117 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file · informational: not an instruction file · information
- `prompts/nightly-review.md:101` · Instructs the agent to modify its persistent memory or instruction files — prompts/nightly-review.md:101 · Instructs the agent to modify its persistent memory or instruction files
- `prompts/nightly-review.md:79` · Instructs the agent to modify its persistent memory or instruction files — prompts/nightly-review.md:79 · Instructs the agent to modify its persistent memory or instruction files
- `scripts/trend_detection.py:395` · Instructs the agent to modify its persistent memory or instruction files — scripts/trend_detection.py:395 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file · informational: not an instruction file · informatio
- `prompts/nightly-review.md:77` · Instructs the agent to modify its persistent memory or instruction files — prompts/nightly-review.md:77 · Instructs the agent to modify its persistent memory or instruction files
- `examples/review-sample.md:19` · Instructs the agent to modify its persistent memory or instruction files — examples/review-sample.md:19 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file · informational: not an instruction file · informationa

## Proposed change
The attached patch removes the 6 lines above. If the behavior is intended, the fix is to declare it instead: list the host under `permissions.network` (or the MCP manifest) and say in the README why the package needs it; the next crawl will re-score it with the declaration taken into account.

## How this will be verified
Nothing is "verified safe" by assertion. When a new version is published, KillerSkills re-runs it in the honeypot and the verdict page updates; you can trigger it early from the verdict page ("Re-scan after I publish a fix"). Reply here or at [email protected] if you believe the verdict is wrong; disputes are reviewed by a person and the outcome is published on the verdict page.

_Proposed through killerskills.ai/threats by a user of the package, under their own GitHub account. The evidence is public; the package was never modified by KillerSkills._