jobsearch-workflow-tailor-apply
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches external job postings via URLs or pasted text, which introduces a potential attack surface for indirect prompt injection where an attacker could embed instructions to manipulate the agent's behavior.
- Ingestion points: In the 'Intake' section of
SKILL.md, the agent is instructed to fetch posting content from links or pasted text. - Boundary markers: Explicit delimiters for the ingested external content are not defined in the instructions.
- Capability inventory: The skill has the ability to read sensitive configuration files (
jobsearch-connections.md,job-search-os-profile.md), write local files (.docx, .pdf, .txt), create Gmail drafts, and publish/update artifacts using Claude's storage capability (window.claude.use('db')intemplates/application-board.html). - Sanitization: The skill implements a 'Compliance pass' in
SKILL.mdthat checks for traceability of metrics and accomplishments to the user's profile, providing a verification layer against hallucinated or injected claims in the output.
Audit Metadata