The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted data from the target project's source code and configuration files via MCP tools.
Ingestion points: Project metadata and source code accessed via tools such as get_entity_details, list_project_endpoints, and list_all_domain_entities in SKILL.md.
Boundary markers: The skill does not specify explicit delimiters for the data returned by MCP tools, though it operates within a structured subagent workflow.
Capability inventory: The skill is limited to project exploration and static analysis; it does not contain capabilities for arbitrary command execution or filesystem modification.
Sanitization: The skill implements a robust sanitization step in its subagent instructions, requiring the redaction of any values matching common credential keys (e.g., password, secret, token).
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill references the author's official marketplace at amplicode.ru to guide users in installing necessary prerequisites. This is consistent with the skill's stated purpose and originates from a legitimate vendor resource.
[SAFE]: A strong security control is implemented in the execution phase where the agent is instructed to identify and redact sensitive information (credentials, API keys, private keys) before the exploration report is generated.

spring-explore