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. 68 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 6 caught lines
Download .patchdiff --git a/src/mcpiranha/rules/builtin/source_behavior.py b/src/mcpiranha/rules/builtin/source_behavior.py --- a/src/mcpiranha/rules/builtin/source_behavior.py +++ b/src/mcpiranha/rules/builtin/source_behavior.py @@ -9,1 +9,0 @@ -- SRC-003: literal paths to credential stores (~/.ssh, ~/.aws/credentials, browser logins), in @@ -11,1 +10,0 @@ -tunnels read ~/.ssh/id_rsa, security tools list these paths to block them. @@ -12,1 +10,0 @@ -- SRC-004: running decoded or downloaded code (eval(atob(...)), exec(b64decode(...))). Not diff --git a/src/mcpiranha/core/models.py b/src/mcpiranha/core/models.py --- a/src/mcpiranha/core/models.py +++ b/src/mcpiranha/core/models.py @@ -175,1 +175,0 @@ -"sensitive_file", # a literal path to credentials (~/.ssh, ~/.aws/credentials, ...) diff --git a/src/mcpiranha/rules/builtin/sensitive_paths.py b/src/mcpiranha/rules/builtin/sensitive_paths.py --- a/src/mcpiranha/rules/builtin/sensitive_paths.py +++ b/src/mcpiranha/rules/builtin/sensitive_paths.py @@ -3,1 +3,0 @@ -A calculator or weather tool has no reason to tell the model to read `~/.ssh/id_rsa`, `.env` or the diff --git a/src/mcpiranha/source/taint_python.py b/src/mcpiranha/source/taint_python.py --- a/src/mcpiranha/source/taint_python.py +++ b/src/mcpiranha/source/taint_python.py @@ -11,1 +11,0 @@ -- whole: the argument itself (`exec(command)` in a shell tool, `fetch(url)` in a fetch tool) is
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.4.1. 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: mcpiranha: Python exec or eval of computed code (KillerSkills verdict caution 68)
## 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 68/100**. Summary: no network, touches nothing outside the project, writes no memory. - See the findings below. Full trace and findings: https://killerskills.ai/skills/pypi/mcpiranha/0.4.1 ## Findings with a location - `src/mcpiranha/rules/builtin/source_behavior.py:12` · Python exec or eval of computed code — src/mcpiranha/rules/builtin/source_behavior.py:12 · Python exec or eval of computed code - `src/mcpiranha/core/models.py:175` · References a credential file in the home directory — src/mcpiranha/core/models.py:175 · References a credential file in the home directory - `src/mcpiranha/rules/builtin/source_behavior.py:9` · References a credential file in the home directory — src/mcpiranha/rules/builtin/source_behavior.py:9 · References a credential file in the home directory - `src/mcpiranha/rules/builtin/source_behavior.py:11` · Reads credential paths from the home directory — src/mcpiranha/rules/builtin/source_behavior.py:11 · Reads credential paths from the home directory - `src/mcpiranha/rules/builtin/sensitive_paths.py:3` · Reads credential paths from the home directory — src/mcpiranha/rules/builtin/sensitive_paths.py:3 · Reads credential paths from the home directory - `src/mcpiranha/source/taint_python.py:11` · Python exec or eval of computed code — src/mcpiranha/source/taint_python.py:11 · Python exec or eval of computed code - `src/mcpiranha/rules/builtin/source_behavior.py:12` · Host or code assembled from base64 fragments — src/mcpiranha/rules/builtin/source_behavior.py:12 · Host or code assembled from base64 fragments - `src/mcpiranha/rules/builtin/source_behavior.py:11` · References a credential file in the home directory — src/mcpiranha/rules/builtin/source_behavior.py:11 · References a credential file in the home directory - `src/mcpiranha/rules/builtin/sensitive_paths.py:3` · References a credential file in the home directory — src/mcpiranha/rules/builtin/sensitive_paths.py:3 · References a credential file in the home directory ## 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._
# or from a terminal, under your own account: gh repo fork ethanroubal/MCPiranha --clone && cd MCPiranha curl -s https://killerskills.ai/skills/pypi/mcpiranha/0.4.1/patch.diff | git apply --unidiff-zero git checkout -b killerskills/0.4.1 && git commit -am "mcpiranha: Python exec or eval of computed code" gh pr create --fill --body-file <(curl -s https://killerskills.ai/skills/pypi/mcpiranha/0.4.1/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.