The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as bootstrap and ./api.sh to perform security testing tasks and initialize tool environments.
[EXTERNAL_DOWNLOADS]: The documentation provides links to various external tools hosted on GitHub (e.g., praetorian-inc/vespasian, arjinexe/entropy-chaos, Sharon-Needles/api, ticarpi/jwt_tool) and encourages their installation via pip or direct script execution.
[DATA_EXFILTRATION]: By design, the skill uses tools to send data and requests to external API endpoints for testing purposes. It also accesses files outside the skill's base directory (e.g., ../field-journal/precedent-pentest.md and ../tool-index.md), which could lead to unauthorized data access if the environment contains sensitive logs or credentials.
[REMOTE_CODE_EXECUTION]: The skill's workflow involves downloading scripts from remote repositories and executing them locally to facilitate security assessments, posing a risk if those third-party sources are compromised.
[PROMPT_INJECTION]: The skill processes untrusted external data such as API specifications, robots.txt files, and JavaScript source code to discover endpoints and generate attack scenarios, creating a surface for indirect prompt injection.
Ingestion points: Reads external API specifications (api-spec.yaml), robots.txt, and JS files as described in Phase 1 of the testing methodology.
Boundary markers: Absent. The skill does not define delimiters or warnings to ignore instructions embedded in the ingested data.
Capability inventory: Includes network operations (curl, wget, etc.), local script execution (api.sh, bootstrap), and file writing (vespasian export).
Sanitization: No explicit sanitization or validation of external content is described before processing.

api-security