handle-linkedin-connection-request-signal
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external input from LinkedIn connection request invitation notes and instructs the agent to pass this content verbatim to downstream systems.
- Ingestion points: The
invitation note(referenced in SKILL.md under Payload shape) is an attacker-controlled external input from LinkedIn. - Boundary markers: The skill lacks defined delimiters or "ignore instructions" wrappers when interpolating the invitation note into CRM records, Slack messages, or lead scoring contexts.
- Capability inventory: The skill possesses significant capabilities including writing to the CRM (via
{{CRM_HYGIENE_SKILL}}in Step 6), posting to Slack channels ({{VISIBILITY_CHANNEL}}in Step 8), and triggering scoring workflows ({{LEAD_SCORING_SKILL}}in Step 7). - Sanitization: The instructions explicitly mandate that the note be carried through "verbatim" and "never summarized," which removes opportunities for the agent to sanitize or neutralize potentially malicious instructions embedded in the note.
Audit Metadata