Propose a patch

Evidence

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

diff --git a/apex.md b/apex.md
--- a/apex.md
+++ b/apex.md
@@ -116,1 +116,0 @@
-…AsyncApexJob` | **When a custom endpoint, action or flow enters the picture**: add its row to `## Integrations` in `memory.md` (name, direction, mechanism, objects, owner) and put its contract — path, payl…
diff --git a/limits.md b/limits.md
--- a/limits.md
+++ b/limits.md
@@ -74,1 +74,0 @@
-…completion. Check `DataStorageMB` before a load that adds millions of rows, and write the reading into `## Limits Observed` in `memory.md`. ## Budgeting a Job State this before running anything large: ``` calls = c…
diff --git a/soql.md b/soql.md
--- a/soql.md
+++ b/soql.md
@@ -172,1 +172,0 @@
-…he user asks for repeatedly, or it replaces a report that hit its row ceiling — add a row to `## Saved Queries` in `~/Clawic/data/salesforce-api-integration/memory.md` with its purpose and whether the filter is selective. If a query's failure tau…
diff --git a/composite.md b/composite.md
--- a/composite.md
+++ b/composite.md
@@ -126,1 +126,0 @@
-…it — save it as `artifacts/<kebab-name>.md` with a line on when to read it, and add its `## Boxes` line in `memory.md` the same turn. Rebuilding a 25-step chain from scratch costs an afternoon; rea…
diff --git a/objects.md b/objects.md
--- a/objects.md
+++ b/objects.md
@@ -130,1 +130,0 @@
-…es, a picklist restricted by record type, a master-detail you cannot reparent — add or update its row in `## Schema Map` in `memory.md`. If you had to derive the whole field list, it goes to `schema/<object-api-nam…

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.0.3. 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: salesforce-api-integration: Instructs the agent to modify its persistent memory or instruction files (KillerSkills verdict caution 69)

## 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 69/100**. Summary: static analysis only — behavior not yet observed.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/clawhub/salesforce-api-integration/1.0.3

## Findings with a location
- `apex.md:116` · Instructs the agent to modify its persistent memory or instruction files — apex.md:116 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `limits.md:74` · Instructs the agent to modify its persistent memory or instruction files — limits.md:74 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `soql.md:172` · Instructs the agent to modify its persistent memory or instruction files — soql.md:172 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `composite.md:126` · Instructs the agent to modify its persistent memory or instruction files — composite.md:126 · Instructs the agent to modify its persistent memory or instruction files · informational: not an instruction file
- `objects.md:130` · Instructs the agent to modify its persistent memory or instruction files — objects.md:130 · 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._