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. 64 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 12 caught lines
Download .patchdiff --git a/memory/lessons.md b/memory/lessons.md --- a/memory/lessons.md +++ b/memory/lessons.md @@ -1,1 +1,0 @@ -memory/lessons.md - Lessons Learned WHAT THIS IS: Hard-won lessons from real production use. These are not theoretical — each one represents a real failure, rea @@ -40,1 +39,0 @@ -**The rule:** If you want to remember something across sessions, write it to a file. `MEMORY.md`, `memory/YYYY-MM-DD.md`, `TOOLS.md` - pick the right home. No exceptions. diff --git a/FIRST-SESSION.md b/FIRST-SESSION.md --- a/FIRST-SESSION.md +++ b/FIRST-SESSION.md @@ -148,1 +148,0 @@ -3. **Log it:** Add a line to MEMORY.md: `"First session incomplete - X questions remaining (see calibration-state.json)."` @@ -177,1 +176,0 @@ -- Update memory files (MEMORY.md, daily logs, USER.md) @@ -254,1 +252,0 @@ -Write any personal additions into `AGENTS.md`. Copy the key hard limits to USER.md under "Never do without asking" so both files stay consistent. @@ -277,1 +274,0 @@ -Then write a seed entry to `MEMORY.md` - one or two lines capturing the most important things from this session. The stuff that should survive a context reset. diff --git a/AGENTS.md b/AGENTS.md --- a/AGENTS.md +++ b/AGENTS.md @@ -95,1 +95,0 @@ -- You can **read, edit, and update** MEMORY.md freely in main sessions @@ -98,1 +97,0 @@ -- Over time, review your daily files and update MEMORY.md with what's worth keeping @@ -105,1 +103,0 @@ -- When you learn a lesson → update AGENTS.md, TOOLS.md, or the relevant skill @@ -334,1 +331,0 @@ -- **Review and update MEMORY.md** (see below) diff --git a/TOOLS.md b/TOOLS.md --- a/TOOLS.md +++ b/TOOLS.md @@ -1,1 +1,0 @@ -TOOLS.md - Local Tool Notes WHAT THIS IS: Your agent's cheat sheet for YOUR specific setup. Skills define how tools work. This file is for the stuff unique to y diff --git a/MEMORY.md b/MEMORY.md --- a/MEMORY.md +++ b/MEMORY.md @@ -9,1 +9,0 @@ -- You can also tell your agent "update MEMORY.md with X" at any time
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.0.1. 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: scaffold: HTML comment carrying imperative instructions to the agent (KillerSkills verdict caution 64)
## 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 64/100**. Summary: static analysis only — behavior not yet observed. - See the findings below. Full trace and findings: https://killerskills.ai/skills/clawhub/scaffold/1.0.1 ## Findings with a location - `memory/lessons.md:1` · HTML comment carrying imperative instructions to the agent — memory/lessons.md:1 · HTML comment carrying imperative instructions to the agent - `FIRST-SESSION.md:254` · Instructs the agent to modify its persistent memory or instruction files — FIRST-SESSION.md:254 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `memory/lessons.md:40` · Instructs the agent to modify its persistent memory or instruction files — memory/lessons.md:40 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `FIRST-SESSION.md:277` · Instructs the agent to modify its persistent memory or instruction files — FIRST-SESSION.md:277 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `AGENTS.md:98` · Instructs the agent to modify its persistent memory or instruction files — AGENTS.md:98 · Instructs the agent to modify its persistent memory or instruction files - `AGENTS.md:105` · Instructs the agent to modify its persistent memory or instruction files — AGENTS.md:105 · Instructs the agent to modify its persistent memory or instruction files - `TOOLS.md:1` · HTML comment carrying imperative instructions to the agent — TOOLS.md:1 · HTML comment carrying imperative instructions to the agent - `AGENTS.md:95` · Instructs the agent to modify its persistent memory or instruction files — AGENTS.md:95 · Instructs the agent to modify its persistent memory or instruction files - `AGENTS.md:334` · Instructs the agent to modify its persistent memory or instruction files — AGENTS.md:334 · Instructs the agent to modify its persistent memory or instruction files - `FIRST-SESSION.md:148` · Instructs the agent to modify its persistent memory or instruction files — FIRST-SESSION.md:148 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `FIRST-SESSION.md:177` · Instructs the agent to modify its persistent memory or instruction files — FIRST-SESSION.md:177 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `MEMORY.md:9` · Instructs the agent to modify its persistent memory or instruction files — MEMORY.md:9 · Instructs the agent to modify its persistent memory or instruction files ## Proposed change The attached patch removes the 12 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._
# or from a terminal, under your own account: gh repo fork scaffoldworkspace/scaffold --clone && cd scaffold curl -s https://killerskills.ai/skills/clawhub/scaffold/1.0.1/patch.diff | git apply --unidiff-zero git checkout -b killerskills/1.0.1 && git commit -am "scaffold: HTML comment carrying imperative instructions to the agent" gh pr create --fill --body-file <(curl -s https://killerskills.ai/skills/clawhub/scaffold/1.0.1/patch.md)
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.