The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides setup instructions that involve downloading and executing scripts directly from the vendor's website via the command line.
Evidence: The command curl -fsSL https://geoly.ai/install.sh | sh is recommended for macOS/Linux installation in SKILL.md.
Evidence: The command powershell -ExecutionPolicy Bypass -c "irm https://geoly.ai/install.ps1 | iex" is recommended for Windows installation in SKILL.md.
[PRIVILEGE_ESCALATION]: To install the CLI on Windows, the skill instructs the user to bypass the PowerShell execution policy, which can allow the execution of unsigned or malicious scripts.
Evidence: Use of the -ExecutionPolicy Bypass flag in the installation command within SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill processes summaries and answers from external AI engines, which are inherently untrusted and can be used to inject malicious instructions.
Ingestion points: Data is ingested through tools like list_brand_answers and get_prompt_record_detail as described in SKILL.md.
Boundary markers: There are no specific boundary markers or instructions provided to the agent to treat this external data as untrusted.
Capability inventory: The skill has the capability to execute shell commands via the geoly CLI and modify configurations via write tools (e.g., create_prompt, archive_prompt).
Sanitization: No sanitization or verification processes are documented for the ingested content.
[COMMAND_EXECUTION]: The skill operates by executing various subcommands of the geoly CLI tool.
Evidence: Frequent references to geoly run, geoly call, and other CLI commands in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill fetches a remote configuration file to check for plugin updates.
Evidence: A version check is performed against https://raw.githubusercontent.com/geoly-ai/codex-plugins/main/plugins/geoly-mcp/.codex-plugin/plugin.json.

geoly-mcp