Skip to content

plan.md

Latest commit

 

History

History
47 lines (37 loc) · 2.83 KB

plan.md

File metadata and controls

47 lines (37 loc) · 2.83 KB
Raw

Zypheron CLI C2 + CLI Polish Plan

Summary

Finish the current C2 integration and CLI polish work as a scoped hardening pass. The goal is to make C2 operations operator-controlled, improve shell/JSON UX, keep installer guidance aligned with the docs, and verify the changed behavior with focused tests plus broad suite runs.

Implementation Scope

  • Add guided C2 workflow support through zypheron exploit --c2 <framework> --guided.
  • Keep C2 actions explicit: listing sessions/agents is the default, listener creation requires --listener, and guided mode must not generate payloads or run autonomous exploitation.
  • Support C2 completions for --c2 and --listener.
  • Ensure Metasploit C2 execution uses msfconsole.
  • Validate listener types before invoking external C2 tooling.
  • Add zypheron completion bash|zsh|fish.
  • Make zypheron scan ... --format json non-interactive, non-streaming, and JSON-only on stdout.
  • Keep optional C2 frameworks out of bulk installs and point Sliver/Empire users to scripts/install/install-c2.sh.
  • Update live docs: README.md, docs/CLI_REFERENCE.md, and docs/INSTALL.md.

Review Focus

  • C2 safety boundary: guided mode must remain setup-only unless the operator also supplies an explicit action such as --listener.
  • C2 executable routing: Sliver/Havoc use their framework binary, Empire uses REST API env vars, and Metasploit uses msfconsole.
  • Machine-readable output: JSON scan mode must not print banners, spinners, warnings, styled text, or storage notices to stdout.
  • Backward compatibility: normal scan text output, existing report generation, and existing tool detection should continue to work.
  • Documentation consistency: CLI examples and installer guidance should match implemented flags and paths.

Test Plan

  • Run focused Go tests:
    • go test ./internal/commands ./internal/c2/empire ./internal/tools
  • Run broad Go tests:
    • go test ./...
    • Use workspace-local GOCACHE, GOTMPDIR, and CCACHE_DIR when /home cache is read-only or /tmp is mounted noexec.
  • Run Python tests where dependencies are available:
    • zypheron-ai: python -m pytest tests -q -o addopts=''
    • zypheron-api: python -m pytest tests -q
  • Record environment-blocked failures separately from product failures. Known blockers in this sandbox:
    • Go internal/desktopbridge tests require local socket listeners.
    • Some Python envs may lack dependencies such as pytest, loguru, networkx, or stripe.

Acceptance Criteria

  • Focused Go tests for changed packages pass.
  • Broad Go suite has no failures except documented sandbox listener restrictions.
  • Python suites either pass in their virtualenvs or fail only on missing local dependencies.
  • No generated cache, coverage, or temp artifacts remain in the working tree.
  • Review findings are either fixed or documented with file/line references.