Skip to content

claude-code.md

Latest commit

 

History

History
49 lines (34 loc) · 1.77 KB

claude-code.md

File metadata and controls

49 lines (34 loc) · 1.77 KB
Raw

Runtime: Claude Code

Claude Code is a terminal agent with native MCP and the same agentskills.io skill format f0_sectools uses — so our skills work unmodified. (Note: Claude Code runs Anthropic's hosted models, not a local model. Use it for development and for driving the tools where local-only privacy isn't required; for fully-local operation use Hermes or LM Studio.)

Prerequisite: finish getting started.

Add the MCP servers

From the repo root:

claude mcp add f0-defender -- uv run --directory "$(pwd)" f0-defender-mcp
claude mcp add f0-entra    -- uv run --directory "$(pwd)" f0-entra-mcp

Or add a project-scoped .mcp.json (same mcpServers format as examples/mcp/mcp.json).

Tools appear as mcp__f0-defender__list_incidents, etc.

Use the skills

Claude Code discovers skills from .claude/skills/ (project) or ~/.claude/skills/. Make our skills available, e.g. symlink them:

mkdir -p .claude/skills
ln -s "$(pwd)/skills/defender/triage-incident"  .claude/skills/triage-defender-incident
ln -s "$(pwd)/skills/defender/posture-summary"  .claude/skills/defender-posture-summary
ln -s "$(pwd)/skills/defender/threat-hunt"      .claude/skills/defender-threat-hunt

Then ask naturally ("triage our active incidents", "posture summary") and the matching skill activates.

Personas

Claude Code has no /personality switch. For a role lens, paste the relevant mode from prompts/f0-sectools-system-prompt.md into your request, or add it to a project CLAUDE.md.

Notes

  • Everything is read-only; no gated write actions are exposed yet.
  • Skills are the same files used by Hermes — no Claude-specific copies.