The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill explicitly instructs the agent to access and use the linkedin_li_at session cookie from the user's configuration (userConfig.linkedin_li_at). This cookie provides full authenticated access to the user's LinkedIn account.
[DATA_EXFILTRATION]: The sensitive LinkedIn session cookie is passed to external third-party services (Apify) to facilitate scraping. This creates a high risk of credential exfiltration if the target scripts are malicious or become compromised.
[EXTERNAL_DOWNLOADS]: The skill relies on several community-contributed scripts (Apify actors) whose source code is managed by third parties: scrapemint/linkedin-company-employees-scraper, harvestapi/linkedin-company-employees, and tri_angle/similarweb-scraper. These represent unverifiable external dependencies.
[REMOTE_CODE_EXECUTION]: By instructing the agent to invoke specific third-party Apify actors, the skill triggers the execution of remote code on the Apify platform.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to scrape and extract "verbatim" phrases from arbitrary external websites.
Ingestion points: External IT agency websites provided as arguments.
Boundary markers: None. The skill does not provide instructions to use delimiters or ignore embedded instructions in the scraped content.
Capability inventory: The skill has significant capabilities, including network access (WebFetch/Apify) and the ability to write data to a persistent Notion page.
Sanitization: Absent. There is no evidence of filtering or escaping logic for the scraped data, allowing malicious instructions on a target website to potentially influence the agent's logic or poison the resulting Notion database.

gtm-intake