Propose a patch

Evidence

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

diff --git a/mcppt/checks.py b/mcppt/checks.py
--- a/mcppt/checks.py
+++ b/mcppt/checks.py
@@ -1060,1 +1060,0 @@
-"​": "U+200B Zero Width Space",
@@ -1061,1 +1060,0 @@
-"‌": "U+200C Zero Width Non-Joiner",
@@ -1062,1 +1060,0 @@
-"‍": "U+200D Zero Width Joiner",
@@ -1063,1 +1060,0 @@
-"": "U+FEFF BOM/Zero Width No-Break Space",
diff --git a/docs/MCPTROTTER_MEDIUM.md b/docs/MCPTROTTER_MEDIUM.md
--- a/docs/MCPTROTTER_MEDIUM.md
+++ b/docs/MCPTROTTER_MEDIUM.md
@@ -213,1 +213,0 @@
-**Prompt injection:** Sanitize all tool outputs before they enter LLM context. Strip `<system>`, `</tool_result>`, and template markers. Apply output guardrails that detect instruction-like patterns.
diff --git a/docs/MCPTROTTER_ARTICLE.md b/docs/MCPTROTTER_ARTICLE.md
--- a/docs/MCPTROTTER_ARTICLE.md
+++ b/docs/MCPTROTTER_ARTICLE.md
@@ -267,1 +267,0 @@
-- Escape or strip special sequences (`<system>`, `</tool_result>`, `{{`, `}}`) from all tool outputs before they reach the model
diff --git a/vuln_server.py b/vuln_server.py
--- a/vuln_server.py
+++ b/vuln_server.py
@@ -201,1 +201,0 @@
-"​‌‍"  # Zero-width Space, Non-Joiner, Joiner (invisible)

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 3.1.0. 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: mcppt: Hidden zero-width characters detected (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/pypi/mcppt/3.1.0

## Findings with a location
- `mcppt/checks.py:1062` · Hidden zero-width characters detected — mcppt/checks.py:1062 · Hidden zero-width characters detected — may conceal content
- `mcppt/checks.py:1063` · Hidden zero-width characters detected — mcppt/checks.py:1063 · Hidden zero-width characters detected — may conceal content
- `mcppt/checks.py:1061` · Hidden zero-width characters detected — mcppt/checks.py:1061 · Hidden zero-width characters detected — may conceal content
- `mcppt/checks.py:1060` · Hidden zero-width characters detected — mcppt/checks.py:1060 · Hidden zero-width characters detected — may conceal content
- `docs/MCPTROTTER_MEDIUM.md:213` · Fake instruction boundary tag — docs/MCPTROTTER_MEDIUM.md:213 · Fake instruction boundary tag · informational: not an instruction file
- `docs/MCPTROTTER_ARTICLE.md:267` · Fake instruction boundary tag — docs/MCPTROTTER_ARTICLE.md:267 · Fake instruction boundary tag · informational: not an instruction file
- `vuln_server.py:201` · Hidden zero-width characters detected — vuln_server.py:201 · Hidden zero-width characters detected — may conceal content

## Proposed change
The attached patch removes the 7 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._