Propose a patch

Evidence

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

diff --git a/debugging.md b/debugging.md
--- a/debugging.md
+++ b/debugging.md
@@ -143,1 +143,0 @@
-…h reason it is shaped that way, with every secret replaced by its pointer — and add its `## Boxes` line in `memory.md` in the same turn (`memory-template.md`). Record the repo's debug entry point a…
diff --git a/SKILL.md b/SKILL.md
--- a/SKILL.md
+++ b/SKILL.md
@@ -40,1 +40,0 @@
-…`env:GITHUB_TOKEN`, `keychain:npm-publish`, `1password:Work/Registry/ci`, `file:~/.ssh/id_ed25519`. If data sits at an old location (`~/vscode/` or `~/clawic/vscode/`), m…
diff --git a/terminal.md b/terminal.md
--- a/terminal.md
+++ b/terminal.md
@@ -106,1 +106,0 @@
-…A needed by extension installs, the profile flags that produce a usable shell — write it as a line in `## Environment` of `~/Clawic/data/vscode/memory.md` in the same turn (`memory-template.md`). If the fix was a terminal profile blo…
diff --git a/extensions.md b/extensions.md
--- a/extensions.md
+++ b/extensions.md
@@ -79,1 +79,0 @@
-…cific Version…** pins an extension after a bad release. The pin holds until you update manually — record why in the extension's `## Extensions` row in `memory.md`, or the pin becomes a mystery. - **Pre-release versions** are a separate chann…
@@ -108,1 +107,0 @@
-…ension verdict** — adopted, dropped, banned, pinned, or blamed for a conflict — write the row in `## Extensions` of `~/Clawic/data/vscode/memory.md` with the id, the verdict and the one-clause reason (`memory-template.md`). A b…

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.0.2. 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: vscode: Instructs the agent to modify its persistent memory or instruction files (KillerSkills verdict caution 65)

## 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 65/100**. Summary: static analysis only — behavior not yet observed.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/clawhub/vscode/1.0.2

## Findings with a location
- `debugging.md:143` · Instructs the agent to modify its persistent memory or instruction files — debugging.md:143 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `SKILL.md:40` · References a credential file in the home directory — SKILL.md:40 · References a credential file in the home directory
- `terminal.md:106` · Instructs the agent to modify its persistent memory or instruction files — terminal.md:106 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `extensions.md:108` · Instructs the agent to modify its persistent memory or instruction files — extensions.md:108 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `extensions.md:79` · Instructs the agent to modify its persistent memory or instruction files — extensions.md:79 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file

## Proposed change
The attached patch removes the 5 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._