The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill clones the Shannon framework repository from 'github.com/KeygraphHQ/shannon' to the local file system (defaulting to '~/shannon'). This is the intended primary function of the skill to provide the underlying security toolset.
[COMMAND_EXECUTION]: The skill uses the Bash tool to manage the framework's lifecycle, including git operations (clone/pull), directory creation, symlinking source code, and executing the Shannon CLI ('./shannon start'). All significant operations are performed within Docker containers as managed by the tool.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to ingest and analyze untrusted third-party source code and interact with external target URLs.
Ingestion points: Target application source code (via 'REPO_PATH') and live application endpoints (via 'TARGET_URL').
Boundary markers: The skill does not implement specific delimiters for ingested code but relies on its multi-agent architecture and Docker isolation.
Capability inventory: The skill has access to Bash for local orchestration and comprehensive network/exploitation tools within its Docker containers.
Sanitization: No explicit sanitization of source code is performed; however, the skill mandates a 'Safety Checks' phase where the user must confirm authorization and the non-production status of the target before execution.
[DATA_EXFILTRATION]: The skill checks for environment variables containing AI provider credentials (such as 'ANTHROPIC_API_KEY'). These are used legitimately by the Shannon framework to facilitate its autonomous reasoning process and are not observed being sent to unauthorized third-party domains.

shannon