Propose a patch

Evidence

The sandbox observed nothing beyond the static findings; the proposal cites those.

diff --git a/tests/test_final_review.py b/tests/test_final_review.py
--- a/tests/test_final_review.py
+++ b/tests/test_final_review.py
@@ -25,1 +25,0 @@
-f"echo {CURL_B64} | base64 -d | bash",
@@ -26,1 +25,0 @@
-f"echo {CURL_B64} | base64 --decode | sh",
@@ -27,1 +25,0 @@
-f"base64 -d payload.b64 | bash",
@@ -29,1 +26,0 @@
-f"bash -c \"$(echo {CURL_B64} | base64 -d)\"",
@@ -30,1 +26,0 @@
-f"eval \"$(echo {CURL_B64} | base64 -d)\"",
diff --git a/src/scamgate/hook.py b/src/scamgate/hook.py
--- a/src/scamgate/hook.py
+++ b/src/scamgate/hook.py
@@ -611,1 +611,0 @@
-{_DECODE}[^\n;&]*\|\s*(?:sudo\s+)?(?:{_RUNNER})\b(?!\s+-?\w*\.\w)           # ... | base64 -d | bash
diff --git a/tests/test_redact.py b/tests/test_redact.py
--- a/tests/test_redact.py
+++ b/tests/test_redact.py
@@ -5,1 +5,0 @@
-r = redact_url("https://bob:[email protected]/pay?token=abc123&x=1#frag")
diff --git a/tests/test_url_rules.py b/tests/test_url_rules.py
--- a/tests/test_url_rules.py
+++ b/tests/test_url_rules.py
@@ -99,1 +99,0 @@
-"https://paypal.com:[email protected]/",        # with a password part
@@ -115,1 +114,0 @@
-v = check_url("https://paypal.com:[email protected]/")
diff --git a/tests/test_hook.py b/tests/test_hook.py
--- a/tests/test_hook.py
+++ b/tests/test_hook.py
@@ -204,1 +204,0 @@
-url = "https://someone:[email protected]/pay?token=abcdef123456&card=4111111111111111"

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.3.3. 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: scamgate: Host or command assembled from base64 fragments (KillerSkills verdict caution 68)

## What the honeypot observed
KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.8.1) 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/glama/scamgate/0.3.3

## Findings with a location
- `tests/test_final_review.py:27` · Host or command assembled from base64 fragments — tests/test_final_review.py:27 · Host or command assembled from base64 fragments
- `tests/test_final_review.py:29` · Host or command assembled from base64 fragments — tests/test_final_review.py:29 · Host or command assembled from base64 fragments
- `tests/test_final_review.py:30` · Host or command assembled from base64 fragments — tests/test_final_review.py:30 · Host or command assembled from base64 fragments
- `tests/test_final_review.py:25` · Host or command assembled from base64 fragments — tests/test_final_review.py:25 · Host or command assembled from base64 fragments
- `src/scamgate/hook.py:611` · Host or command assembled from base64 fragments — src/scamgate/hook.py:611 · Host or command assembled from base64 fragments
- `tests/test_final_review.py:26` · Host or command assembled from base64 fragments — tests/test_final_review.py:26 · Host or command assembled from base64 fragments
- `tests/test_redact.py:5` · URL with embedded credentials — tests/test_redact.py:5 · URL with embedded credentials — may be phishing or credential leaking
- `tests/test_url_rules.py:99` · URL with embedded credentials — tests/test_url_rules.py:99 · URL with embedded credentials — may be phishing or credential leaking
- `tests/test_hook.py:204` · URL with embedded credentials — tests/test_hook.py:204 · URL with embedded credentials — may be phishing or credential leaking
- `tests/test_url_rules.py:115` · URL with embedded credentials — tests/test_url_rules.py:115 · URL with embedded credentials — may be phishing or credential leaking

## 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._