The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The documentation instructs users to use sudo to move a downloaded binary to a system directory, which is a privilege escalation pattern that requires administrative rights.
[REMOTE_CODE_EXECUTION]: The skill suggests an installation method for Ollama that pipes a remote script directly to the shell (curl -fsSL https://ollama.com/install.sh | sh), a high-risk pattern for executing code from external servers.
[EXTERNAL_DOWNLOADS]: The framework fetches binaries and scripts from various external repositories including GitHub and the Ollama domain during installation and troubleshooting.
[COMMAND_EXECUTION]: The framework relies on executing several external CLI tools such as holehe, sherlock, and sublist3r as subprocesses to perform its OSINT functions.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The framework ingests search results, breach records, and social metadata from 11 different external OSINT sources through various tool wrappers.
Boundary markers: There are no explicit delimiters or instructions provided to separate the untrusted data from the agent's core instruction set.
Capability inventory: The skill has the capability to execute shell commands, perform network requests, and write report files, creating an exploitable surface if the ingested data contains malicious instructions.
Sanitization: The skill description provides no information on how the data retrieved from external platforms is sanitized or validated before being processed by the AI agent.

openosint-ai-osint-framework