Propose a patch

Evidence

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

diff --git a/src/lib/schemas.ts b/src/lib/schemas.ts
--- a/src/lib/schemas.ts
+++ b/src/lib/schemas.ts
@@ -6,1 +6,0 @@
-* Inputs comunes a las herramientas. `projectRoot` es la raíz del servicio
@@ -12,1 +11,0 @@
-.describe('Ruta absoluta de la raíz del servicio NestJS destino donde se escriben los archivos.')
@@ -129,1 +127,0 @@
-…ia del proyecto (ver infraRoot); el agente aún debe conciliar los NOMBRES de símbolo (p.ej. ResponseJsonApiBase vs ResponseJsonApi del proyecto).' ) export …
diff --git a/src/lib/learnings.ts b/src/lib/learnings.ts
--- a/src/lib/learnings.ts
+++ b/src/lib/learnings.ts
@@ -314,1 +314,0 @@
-// The answers the user can see (and edit) in AGENTS.md win over the ones saved in the learnings file.
diff --git a/test/refactorAdopt.test.mjs b/test/refactorAdopt.test.mjs
--- a/test/refactorAdopt.test.mjs
+++ b/test/refactorAdopt.test.mjs
@@ -217,1 +217,0 @@
-write(root, 'AGENTS.md', '<!-- BEGIN midd-nestjs-preferences (managed) -->\nAUTHORIZATION: required   # hint\nPOSTMAN: no\n<!-- END midd-nestjs-preferences -->\n')
diff --git a/src/tools/learnings.ts b/src/tools/learnings.ts
--- a/src/tools/learnings.ts
+++ b/src/tools/learnings.ts
@@ -14,1 +14,0 @@
-// What the user wrote in AGENTS.md wins over what was saved in the learnings file: it is the one he can edit.
diff --git a/src/lib/preferences.ts b/src/lib/preferences.ts
--- a/src/lib/preferences.ts
+++ b/src/lib/preferences.ts
@@ -91,1 +91,0 @@
-* that has no block is left alone: the user did not want the MCP to write to AGENTS.md.
diff --git a/test/agentic.test.mjs b/test/agentic.test.mjs
--- a/test/agentic.test.mjs
+++ b/test/agentic.test.mjs
@@ -218,1 +218,0 @@
-write(root, 'AGENTS.md', mine)
@@ -226,1 +225,0 @@
-write(root, 'AGENTS.md', withBlock + after)
@@ -243,1 +241,0 @@
-write(root, 'AGENTS.md', `${head}${AGENTS_BEGIN}\nseccion vieja v0.0.1\n${AGENTS_END}${tail}`)
@@ -259,1 +256,0 @@
-write(root, 'AGENTS.md', '# Proyecto\r\n\r\nlinea con CRLF\r\n')
@@ -273,1 +269,0 @@
-write(root, 'AGENTS.md', broken)

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.1.0. 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: midd-nestjs-hexagonal-architecture-mcp: Hidden zero-width characters detected (KillerSkills verdict caution 71)

## 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 71/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/glama/midd-nestjs-hexagonal-architecture-mcp/0.1.0

## Findings with a location
- `src/lib/schemas.ts:12` · Hidden zero-width characters detected — src/lib/schemas.ts:12 · Hidden zero-width characters detected — may conceal content
- `src/lib/schemas.ts:129` · Hidden zero-width characters detected — src/lib/schemas.ts:129 · Hidden zero-width characters detected — may conceal content
- `src/lib/schemas.ts:6` · Hidden zero-width characters detected — src/lib/schemas.ts:6 · Hidden zero-width characters detected — may conceal content
- `src/lib/learnings.ts:314` · Instructs the agent to modify its persistent memory or instruction files — src/lib/learnings.ts:314 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/refactorAdopt.test.mjs:217` · Instructs the agent to modify its persistent memory or instruction files — test/refactorAdopt.test.mjs:217 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `src/tools/learnings.ts:14` · Instructs the agent to modify its persistent memory or instruction files — src/tools/learnings.ts:14 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `src/lib/preferences.ts:91` · Instructs the agent to modify its persistent memory or instruction files — src/lib/preferences.ts:91 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/agentic.test.mjs:243` · Instructs the agent to modify its persistent memory or instruction files — test/agentic.test.mjs:243 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/agentic.test.mjs:218` · Instructs the agent to modify its persistent memory or instruction files — test/agentic.test.mjs:218 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/agentic.test.mjs:273` · Instructs the agent to modify its persistent memory or instruction files — test/agentic.test.mjs:273 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/agentic.test.mjs:226` · Instructs the agent to modify its persistent memory or instruction files — test/agentic.test.mjs:226 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `test/agentic.test.mjs:259` · Instructs the agent to modify its persistent memory or instruction files — test/agentic.test.mjs:259 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file

## Proposed change
The attached patch removes the 12 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._