Skip to content

plugin.md

Latest commit

 

History

History
56 lines (42 loc) · 1.7 KB

plugin.md

File metadata and controls

56 lines (42 loc) · 1.7 KB
Raw

Plugin & skills

AXguard installs into coding-agent harnesses the same way AwareXone’s Agentic Bug Hunter does: copy skills/ and commands/ into the harness config dir.

Install targets

--agent Global root Project root
claude ~/.claude .claude
cursor ~/.cursor .cursor
opencode ~/.config/opencode .opencode
codex ~/.codex .codex
agents ~/.agents .agents (skills only)
all all of the above — 
./install.sh --agent claude
./install.sh --agent cursor --project
./install.sh --agent all
./uninstall.sh --agent cursor

Manifest: .claude-plugin/plugin.json

Layout expectations

axguard-audit/SKILL.md
axguard-cso/SKILL.md
…
commands/axguard-audit.md
commands/axguard-scan.md
…

Claude Code / Cursor expose commands/*.md as slash commands (/axguard-audit). Skills load by description match or explicit invoke.

Authoring rules

  1. Command — steps an agent can follow without guessing (exact CLI, paths, outputs).
  2. Skill — description must include trigger phrases (“pre-ship”, “audit”, “IDOR”, …).
  3. Keep specialist names stable (Pre-ship Lead, CSO, Triage, …) — they appear in the README catalog.
  4. When adding/removing installable names, update arrays in uninstall.sh.

Local verify

./install.sh --agent cursor --project
# In Cursor/Claude: /axguard-help via asking for axguard audit, or run:
axguard audit . --out-dir .findings/axguard
ls .findings/axguard/

Relationship to CLI

The plugin does not reimplement scanning. It routes the agent to axguard scan / axguard audit and then enforces triage/fix/report discipline.