The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains numerous payloads designed to exploit LDAP search filters, allowing an agent to manipulate backend queries to bypass security controls and authentication mechanisms.
[DYNAMIC_EXECUTION]: Both SKILL.md and references/ldap-exploitation.md include full Python scripts that use the requests library to automate exploitation tasks, such as blind data extraction via character-by-character guessing.
[DATA_EXFILTRATION]: The automated exploitation scripts perform network operations to non-whitelisted external targets (identified by the URL variable) to exfiltrate LDAP attribute values and credentials.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a high risk for indirect injection as it is designed to ingest and process data from external targets to drive its exploitation logic.
Ingestion points: External HTTP responses processed by requests.post and requests.get in the exploitation scripts.
Boundary markers: None identified; responses are evaluated directly for success indicators like "Welcome" strings.
Capability inventory: Network access via the requests library, shell command/payload generation for LDAP injection.
Sanitization: No sanitization or validation of the external target's response content is present before the agent processes it to determine the next extraction step.

ldap-injection