The Agent Skills Directory

[SAFE]: The skill incorporates a mandatory startup gate that requires explicit confirmation of written authorization and a defined scope before any active testing is initiated. This prevents unauthorized scanning activities.
[SAFE]: Instructions are provided to sanitize all external tool outputs by stripping ANSI escape codes and extracting specific fields. This serves as a defensive measure to prevent terminal-based injection or formatting attacks when processing untrusted data.
[COMMAND_EXECUTION]: The workflow directs the agent to execute various security auditing tools (e.g., nmap, nuclei, gobuster) and standard package managers (npx, pip) to facilitate scanning and resolve environment-specific tool issues.
[EXTERNAL_DOWNLOADS]: The skill utilizes vendor-managed resources, recommending the installation of functional extensions like scope-grill and finding-writer using the npx rifteo-skills command.
[SAFE]: The skill implements an evidence chain for handling untrusted data ingested during reconnaissance and scanning:
Ingestion points: Output from various security scanning tools in Phases 1, 2, and 3.
Boundary markers: Explicit triage gates and confidence labeling rules defined in references/triage-rules.md.
Capability inventory: Tool invocations performed via the HexStrike MCP server capabilities.
Sanitization: Required removal of ANSI control characters and metadata from processed strings.

hexstrike-forge