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. 63 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 5 caught lines
Download .patchdiff --git a/template/flydocs/context/project.md b/template/flydocs/context/project.md --- a/template/flydocs/context/project.md +++ b/template/flydocs/context/project.md @@ -31,1 +31,0 @@ -Example: "We use CSS Modules instead of Tailwind for this project" diff --git a/template/flydocs/knowledge/INDEX.md b/template/flydocs/knowledge/INDEX.md --- a/template/flydocs/knowledge/INDEX.md +++ b/template/flydocs/knowledge/INDEX.md @@ -3,1 +3,0 @@ -AGENT INSTRUCTIONS: - Scan this file FIRST when looking for existing knowledge - Use descriptions to assess relevance before loading full docs - Update this ind diff --git a/template/.flydocs/templates/instructions.md b/template/.flydocs/templates/instructions.md --- a/template/.flydocs/templates/instructions.md +++ b/template/.flydocs/templates/instructions.md @@ -111,1 +111,0 @@ -Override specific skill recommendations for this project diff --git a/template/CHANGELOG.md b/template/CHANGELOG.md --- a/template/CHANGELOG.md +++ b/template/CHANGELOG.md @@ -278,1 +278,0 @@ -Run `flydocs update` after upgrading. It brings `AGENTS.md`'s skills index up to @@ -534,1 +533,0 @@ -- `flydocs update` no longer untracks a tracked `AGENTS.md` (or any managed
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.3.45. 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: cli: HTML comment carrying imperative instructions to the agent (KillerSkills verdict caution 63)
## What the honeypot observed KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.7) and scored it **CAUTION 63/100**. Summary: no network, touches nothing outside the project, writes no memory. - See the findings below. Full trace and findings: https://killerskills.ai/skills/npm/cli/1.3.45 ## Findings with a location - `template/flydocs/context/project.md:31` · HTML comment carrying imperative instructions to the agent — template/flydocs/context/project.md:31 · HTML comment carrying imperative instructions to the agent - `template/flydocs/knowledge/INDEX.md:3` · HTML comment carrying imperative instructions to the agent — template/flydocs/knowledge/INDEX.md:3 · HTML comment carrying imperative instructions to the agent - `template/.flydocs/templates/instructions.md:111` · HTML comment carrying imperative instructions to the agent — template/.flydocs/templates/instructions.md:111 · HTML comment carrying imperative instructions to the agent - `template/CHANGELOG.md:278` · Instructs the agent to modify its persistent memory or instruction files — template/CHANGELOG.md:278 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file - `template/CHANGELOG.md:534` · Instructs the agent to modify its persistent memory or instruction files — template/CHANGELOG.md:534 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file ## Proposed change The attached patch removes the 5 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.