service-de-channel-activate
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill uses
node -eto execute inline JavaScript for parsing JSON data retrieved from Salesforce queries. - Evidence: Found in
SKILL.mdandreferences/phone-verification.mdwithin shell loops and status check logic:node -e 'console.log(JSON.parse(require("fs").readFileSync(0,"utf8")).result.records[0].DeploymentStatus)'. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external Salesforce API responses and interpolates it into logic and subsequent commands, creating a potential surface for indirect injection if the source data is compromised.
- Ingestion points: Salesforce record data is read into temporary files such as
/tmp/amc-precheck-channel.jsonand/tmp/amc-poststate-mcu.json(found inSKILL.md). - Boundary markers: Absent; the skill directly parses and uses the JSON properties without explicit delimiters or sanitization instructions for the agent.
- Capability inventory: The skill executes shell commands via the
sfCLI andnode, and performs network writes via RESTPATCHandPOSTrequests. - Sanitization: No explicit sanitization or validation of the ingested JSON fields is performed before they are used in control flow or success reporting.
Audit Metadata