Propose a patch
A fix the maintainer can act on, with the evidence attached. It opens under your GitHub account in your browser; KillerSkills holds no token and never modifies the package. The fix is verified the only way we verify anything: the next crawl re-runs the new version in the honeypot. 66 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 9 caught lines
Download .patchdiff --git a/SKILL.md b/SKILL.md --- a/SKILL.md +++ b/SKILL.md @@ -21,1 +21,0 @@ -- write: MEMORY.md @@ -22,1 +21,0 @@ -- write: CLAUDE.md diff --git a/skill.json b/skill.json --- a/skill.json +++ b/skill.json @@ -13,1 +13,0 @@ -"permissions": ["workspace:read", "workspace:write:.reflect/", "workspace:write:MEMORY.md", "workspace:write:CLAUDE.md"], diff --git a/test/run-eval-test.js b/test/run-eval-test.js --- a/test/run-eval-test.js +++ b/test/run-eval-test.js @@ -58,1 +58,0 @@ -proposed_change: 'Add to MEMORY.md: The openclaw.json config file requires elevated access. Use the OpenClaw CLI or moltbook.py instead of reading the file directly.', @@ -77,1 +76,0 @@ -proposed_change: 'Add to MEMORY.md: On Windows, python3 is a Microsoft Store shim. Use the full path C:/Users/sdysa/AppData/Local/Programs/Python/Python312/python.exe or sys.executable.', @@ -93,1 +91,0 @@ -proposed_change: 'Add to CLAUDE.md: Be more careful when running bash commands. Double check commands before executing.', @@ -112,1 +109,0 @@ -proposed_change: 'Add to MEMORY.md: Always retry connections up to 3 times before failing.', @@ -131,1 +127,0 @@ -proposed_change: 'Add to MEMORY.md: clawhub.io has SSL issues. Use clawhub.ai instead.', diff --git a/assets/evaluator-prompt-binary.md b/assets/evaluator-prompt-binary.md --- a/assets/evaluator-prompt-binary.md +++ b/assets/evaluator-prompt-binary.md @@ -32,1 +32,0 @@ -Proposed change: "Add to MEMORY.md: Use /c/Users/sdysa/AppData/Local/Programs/Python/Python312/python.exe, not python3"
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.0.2. 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: openclaw-reflect: Instructs the agent to modify its persistent memory or instruction files (KillerSkills verdict caution 66)
## What the honeypot observed KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.9.2) and scored it **CAUTION 66/100**. Summary: static analysis only — behavior not yet observed. - See the findings below. Full trace and findings: https://killerskills.ai/skills/clawhub/openclaw-reflect/1.0.2 ## Findings with a location - `SKILL.md:22` · Instructs the agent to modify its persistent memory or instruction files — SKILL.md:22 · Instructs the agent to modify its persistent memory or instruction files - `SKILL.md:21` · Instructs the agent to modify its persistent memory or instruction files — SKILL.md:21 · Instructs the agent to modify its persistent memory or instruction files - `skill.json:13` · Instructs the agent to modify its persistent memory or instruction files — skill.json:13 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `test/run-eval-test.js:112` · Instructs the agent to modify its persistent memory or instruction files — test/run-eval-test.js:112 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `test/run-eval-test.js:77` · Instructs the agent to modify its persistent memory or instruction files — test/run-eval-test.js:77 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `test/run-eval-test.js:131` · Instructs the agent to modify its persistent memory or instruction files — test/run-eval-test.js:131 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `assets/evaluator-prompt-binary.md:32` · Instructs the agent to modify its persistent memory or instruction files — assets/evaluator-prompt-binary.md:32 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `test/run-eval-test.js:58` · Instructs the agent to modify its persistent memory or instruction files — test/run-eval-test.js:58 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `test/run-eval-test.js:93` · Instructs the agent to modify its persistent memory or instruction files — test/run-eval-test.js:93 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file ## Proposed change The attached patch removes the 9 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._
We count proposals opened, not merged: a merge shows up as a new version that the crawl re-runs. Disputes: the maintainer can contest the verdict from its page; a person reviews it and the outcome is published.