The Agent Skills Directory

[COMMAND_EXECUTION]: The skill triggers the execution of a local Python script campaign-tracker.py using dynamic arguments. Specifically, it executes python campaign-tracker.py --brand {slug} --action list-campaigns. This creates a command injection vulnerability if the {slug} variable, which is often derived from user-provided brand names, is not strictly validated.
[DATA_EXFILTRATION]: The skill accesses sensitive brand configuration and profile files located in the ~/.claude-marketing/ hidden directory. It uses dynamic path construction (e.g., ~/.claude-marketing/brands/{slug}/profile.json), which could potentially be exploited for directory traversal if the {slug} variable is manipulated to point to files like ~/.ssh/id_rsa.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of external, attacker-controllable data.
Ingestion points: The skill ingests business names, addresses, and customer reviews during the Local SEO Audit workflow defined in SKILL.md.
Boundary markers: The skill does not employ delimiters or system instructions to ignore embedded commands in the processed data.
Capability inventory: The skill has the capability to read local files and execute shell commands.
Sanitization: There is no evidence of sanitization or escaping of input data before it is processed or used in script parameters.
[REMOTE_CODE_EXECUTION]: The script campaign-tracker.py referenced in SKILL.md is an unverifiable local dependency. Its internal logic and security posture cannot be determined from the provided files, posing a risk if it handles inputs insecurely.

local-seo