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. 76 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 10 caught lines
Download .patchdiff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -30,1 +30,0 @@ -not run: the workflow runs each of these; nesting them here would run the suite inside itself diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -9,1 +9,0 @@ -not run: clones the repository the reader is already in @@ -28,1 +27,0 @@ -not run: these are the gates; running them from inside a gate would recurse @@ -44,1 +42,0 @@ -not run: it reads the owner's own login, which no test may do @@ -115,1 +112,0 @@ -not run: why @@ -184,1 +180,0 @@ -not run: it is already on in this clone diff --git a/docs/setup.md b/docs/setup.md --- a/docs/setup.md +++ b/docs/setup.md @@ -14,1 +14,0 @@ -not run: it installs; the suite runs inside what it installs @@ -68,1 +67,0 @@ -not run: it needs claude-voice and this checkout's path @@ -85,1 +83,0 @@ -not run: it asks for the login @@ -102,1 +99,0 @@ -not run: it asks for typed input
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of r1. 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: easee-mcp: HTML comment carrying imperative instructions to the agent (KillerSkills verdict caution 76)
## What the honeypot observed
KillerSkills ran **easee-mcp@r1** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.9.2) and scored it **CAUTION 76/100**. Summary: no network, touches nothing outside the project, writes no memory.
- See the findings below.
Full trace and findings: https://killerskills.ai/skills/glama/easee-mcp/r1
## Findings with a location
- `README.md:30` · HTML comment carrying imperative instructions to the agent — README.md:30 · HTML comment carrying imperative instructions to the agent
- `CONTRIBUTING.md:28` · HTML comment carrying imperative instructions to the agent — CONTRIBUTING.md:28 · HTML comment carrying imperative instructions to the agent
- `docs/setup.md:14` · HTML comment carrying imperative instructions to the agent — docs/setup.md:14 · HTML comment carrying imperative instructions to the agent
- `docs/setup.md:68` · HTML comment carrying imperative instructions to the agent — docs/setup.md:68 · HTML comment carrying imperative instructions to the agent
- `docs/setup.md:85` · HTML comment carrying imperative instructions to the agent — docs/setup.md:85 · HTML comment carrying imperative instructions to the agent
- `CONTRIBUTING.md:44` · HTML comment carrying imperative instructions to the agent — CONTRIBUTING.md:44 · HTML comment carrying imperative instructions to the agent
- `CONTRIBUTING.md:9` · HTML comment carrying imperative instructions to the agent — CONTRIBUTING.md:9 · HTML comment carrying imperative instructions to the agent
- `docs/setup.md:102` · HTML comment carrying imperative instructions to the agent — docs/setup.md:102 · HTML comment carrying imperative instructions to the agent
- `CONTRIBUTING.md:115` · HTML comment carrying imperative instructions to the agent — CONTRIBUTING.md:115 · HTML comment carrying imperative instructions to the agent
- `CONTRIBUTING.md:184` · HTML comment carrying imperative instructions to the agent — CONTRIBUTING.md:184 · HTML comment carrying imperative instructions to the agent
## Proposed change
The attached patch removes the 10 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 roykollensvendsen/easee-mcp --clone && cd easee-mcp curl -s https://killerskills.ai/skills/glama/easee-mcp/r1/patch.diff | git apply --unidiff-zero git checkout -b killerskills/r1 && git commit -am "easee-mcp: HTML comment carrying imperative instructions to the agent" gh pr create --fill --body-file <(curl -s https://killerskills.ai/skills/glama/easee-mcp/r1/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.