linkedin-post-ideas
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXPOSURE]: The skill reads user configuration files located at
.liftli/voice.mdand~/.liftli/voice.mdto adapt its generated ideas to the user's personal brand and audience. - [EXTERNAL_DOWNLOADS]: The skill instructions recommend that users manually install related tools and MCP servers from the vendor's official domains, including
liftli.aiandmcp.liftli.ai. - [INDIRECT_PROMPT_INJECTION]: The skill processes anecdotes, stories, and expertise provided by the user or stored in local files to generate content ideas. 1. Ingestion points: Reads
.liftli/voice.md,~/.liftli/voice.md, and user responses to interview questions. 2. Boundary markers: No explicit delimiters or 'ignore embedded instructions' markers are used when reading these inputs. 3. Capability inventory: The skill can invoke thestrategist_briefingtool or thelinkedin-publishskill if available in the agent environment. 4. Sanitization: No sanitization or validation of the input data is performed before interpolation into the generation process.
Audit Metadata