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. 69 · Caution
Evidence
The sandbox observed nothing beyond the static findings; the proposal cites those.
Diff · removes 25 caught lines
Download .patchdiff --git a/agents/_build_pr.py b/agents/_build_pr.py
--- a/agents/_build_pr.py
+++ b/agents/_build_pr.py
@@ -129,1 +129,0 @@
-_run(["git", "remote", "set-url", "origin", f"https://x-access-token:{token}@github.com/{REPO_SLUG}.git"])
diff --git a/agents/self_improve.py b/agents/self_improve.py
--- a/agents/self_improve.py
+++ b/agents/self_improve.py
@@ -271,1 +271,0 @@
-_run(["git", "remote", "set-url", "origin", f"https://x-access-token:{token}@github.com/{REPO_SLUG}.git"])
diff --git a/tests/test_code_lint.py b/tests/test_code_lint.py
--- a/tests/test_code_lint.py
+++ b/tests/test_code_lint.py
@@ -20,1 +20,0 @@
-findings = _lint("x.py", "eval('1 + 1')\n")
@@ -175,1 +174,0 @@
-bad.write_text("eval(input())\n")
@@ -195,1 +193,0 @@
-(tmp_path / "bad.py").write_text("eval(input())\n")
diff --git a/agents/redteam_builder.py b/agents/redteam_builder.py
--- a/agents/redteam_builder.py
+++ b/agents/redteam_builder.py
@@ -34,1 +34,0 @@
-"eval(" — a known, documented gap (see
@@ -80,1 +79,0 @@
-"organization avoids the literal substring 'eval(' in source for "
diff --git a/agents/builder.py b/agents/builder.py
--- a/agents/builder.py
+++ b/agents/builder.py
@@ -84,1 +84,0 @@
-"eval(", # arbitrary code execution
@@ -85,1 +84,0 @@
-"exec(", # arbitrary code execution
diff --git a/tests/test_code_review.py b/tests/test_code_review.py
--- a/tests/test_code_review.py
+++ b/tests/test_code_review.py
@@ -117,1 +117,0 @@
-monkeypatch.setattr(cr, "fetch_file_at_ref", lambda path, ref: "eval(user_input)\n")
@@ -154,1 +153,0 @@
-assert cr._rule_tag_for_message("eval() called on a dynamically-built string") == "eval-exec-dynamic"
@@ -216,1 +214,0 @@
-findings = [("HIGH", "agents/bad.py", 3, "eval() called on a dynamically-built string")]
@@ -222,1 +219,0 @@
-findings = [("HIGH", "x.py", 3, "eval() called on a dynamic string")]
@@ -223,1 +219,0 @@
-system, prompt = cr.build_review_prompt("diff --git a/x b/x\n+eval(x)\n", findings)
@@ -224,1 +219,0 @@
-assert "eval() called on a dynamic string" in prompt
@@ -225,1 +219,0 @@
-assert "eval(x)" in prompt
@@ -234,1 +227,0 @@
-_, prompt = cr.build_review_prompt("diff --git a/x b/x\n+eval(x)\n", [])
@@ -264,1 +256,0 @@
-[("HIGH", "x.py", 3, "eval() called on a dynamic string")],
@@ -331,1 +322,0 @@
-fake = _FakeGitHub(files=[{"path": "agents/x.py", "status": "modified", "patch": "+eval(x)"}])
@@ -350,1 +340,0 @@
-("HIGH", "agents/bad.py", 3, "eval() called on a dynamically-built string"),
@@ -361,1 +350,0 @@
-"eval() called on a dynamically-built string [Previously reviewed and confirmed a false "
@@ -363,1 +351,0 @@
-("HIGH", "agents/other.py", 5, "eval() called on a dynamically-built string"),
@@ -421,1 +408,0 @@
-fake._file_contents["agents/bad.py"] = "eval(x)\n"
@@ -441,1 +427,0 @@
-fake._file_contents["agents/bad.py"] = "eval(x)\n"
@@ -457,1 +442,0 @@
-fake._file_contents["agents/bad.py"] = "eval(x)\n"
Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.1.1. 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: vape-mcp-server: URL with embedded credentials (KillerSkills verdict caution 69)
## 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 69/100**. Summary: no network, touches nothing outside the project, writes no memory. - See the findings below. Full trace and findings: https://killerskills.ai/skills/pypi/vape-mcp-server/1.1.1 ## Findings with a location - `agents/_build_pr.py:129` · URL with embedded credentials — agents/_build_pr.py:129 · URL with embedded credentials — may be phishing or credential leaking - `agents/self_improve.py:271` · URL with embedded credentials — agents/self_improve.py:271 · URL with embedded credentials — may be phishing or credential leaking - `tests/test_code_lint.py:20` · Python exec or eval of computed code — tests/test_code_lint.py:20 · Python exec or eval of computed code - `agents/redteam_builder.py:80` · Python exec or eval of computed code — agents/redteam_builder.py:80 · Python exec or eval of computed code - `agents/builder.py:85` · Python exec or eval of computed code — agents/builder.py:85 · Python exec or eval of computed code - `tests/test_code_review.py:222` · Python exec or eval of computed code — tests/test_code_review.py:222 · Python exec or eval of computed code - `tests/test_code_review.py:117` · Python exec or eval of computed code — tests/test_code_review.py:117 · Python exec or eval of computed code - `tests/test_code_review.py:154` · Python exec or eval of computed code — tests/test_code_review.py:154 · Python exec or eval of computed code - `agents/redteam_builder.py:34` · Python exec or eval of computed code — agents/redteam_builder.py:34 · Python exec or eval of computed code - `tests/test_code_review.py:234` · Python exec or eval of computed code — tests/test_code_review.py:234 · Python exec or eval of computed code - `agents/builder.py:84` · Python exec or eval of computed code — agents/builder.py:84 · Python exec or eval of computed code - `tests/test_code_review.py:224` · Python exec or eval of computed code — tests/test_code_review.py:224 · Python exec or eval of computed code ## Proposed change The attached patch removes the 25 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 jUXTAPOSITION1/V.A.P.E --clone && cd V.A.P.E curl -s https://killerskills.ai/skills/pypi/vape-mcp-server/1.1.1/patch.diff | git apply --unidiff-zero git checkout -b killerskills/1.1.1 && git commit -am "vape-mcp-server: URL with embedded credentials" gh pr create --fill --body-file <(curl -s https://killerskills.ai/skills/pypi/vape-mcp-server/1.1.1/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.