The Agent Skills Directory

[COMMAND_EXECUTION]: The workflow in SKILL.md (Steps 2 and 4) instructs the agent to run shell commands that interpolate variables such as ACTOR_ID and JSON_INPUT directly into the command line. This pattern is highly susceptible to command injection if the agent does not properly escape or validate the user-provided inputs, allowing for arbitrary code execution on the host system.
[CREDENTIALS_UNSAFE]: The skill instructs the agent to read a sensitive APIFY_TOKEN from a .env file and use it in shell commands (e.g., export $(grep APIFY_TOKEN .env | xargs)). This exposes the credential in the process environment and potentially in shell history or logs.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external social media platforms (Instagram, TikTok, Facebook, YouTube) via Apify. Since this content is attacker-controlled or untrusted, it could contain malicious instructions designed to manipulate the agent's summary or influence subsequent tool calls.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to api.apify.com to trigger actor runs and download dataset results. This is a documented interaction with the Apify service.
[SAFE_PRACTICE]: The included run_actor.js script implements basic CSV injection mitigation in the csvCell function by escaping values that start with sensitive characters like =, +, @, or -.

apify-influencer-discovery