The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes Amazon customer feedback (reviews and return topics) which are external, user-generated sources. An adversary could potentially embed instructions within a product review to influence the agent's behavior during analysis.
Ingestion points: Retrieved data from SP-API endpoints such as getItemReviewTopics, getBrowseNodeReviewTopics, and getItemReviewTrends (referenced in SKILL.md and scripts/ files).
Boundary markers: The scripts do not explicitly define boundary markers or 'ignore' instructions for the content of the reviews being processed.
Capability inventory: The skill has capabilities to write files to the workspace, access vendor APIs via network requests, and execute local scripts via subprocess.
Sanitization: There is no evidence of explicit sanitization or filtering of the natural language feedback data before it is presented to the agent context.
[PERSISTENCE]: The skill's onboarding documentation (references/onboarding.md) provides instructions for users to append environment variables to shell profile files (~/.zshrc, ~/.bashrc) to store API keys. While intended for legitimate configuration of the CLI environment, this involves modifying shell persistence files.
[COMMAND_EXECUTION]: The script scripts/_spapi_customer_feedback_common.py executes a local dependency checker using subprocess.run([sys.executable, str(checker)]). This is a controlled execution of an internal script to ensure the required authentication skill is available.
[CREDENTIALS_UNSAFE]: The file scripts/onboarding.py contains a hardcoded Base64-encoded identifier _LOGIN_FIXED_UID (eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9). While this appears to be a static client/application ID rather than a private secret, hardcoding identifiers used for authentication is a discouraged practice.
[EXTERNAL_DOWNLOADS]: The skill interacts with external API endpoints at tool-gateway.linkfox.com, api.linkfox.com, and agent-api.linkfox.com. These are verified vendor resources belonging to the skill author (linkfox-ai) and are used for standard operations like authentication proxying and billing.

linkfox-amazon-store-customer-feedback