Use via MCP

Integrate axm-audit into AI agent workflows through the AXM MCP server.

Setup

The axm-audit tool is exposed as audit by the axm-mcp server. No separate installation needed — if axm-mcp is running, the audit tool is available.

Usage

From an AI agent

Call the audit MCP tool with the project path:

{"tool": "audit", "kwargs": {"path": "/path/to/project"}}

The output uses format_agent — compact strings for clean passes, detailed dicts for actionable items:

{
  "score": 85.0,
  "grade": "B",
  "passed": ["QUALITY_LINT: Lint score: 100/100 (0 issues)"],
  "failed": [{"rule_id": "...", "details": {...}, "fix_hint": "..."}]
}

One-shot verification

Use verify instead of audit for a combined quality + governance check:

{"tool": "verify", "kwargs": {"path": "/path/to/project"}}

This runs audit + init_check + AST enrichment in a single call.

Agent-optimized test runner

Use audit_test for structured, token-efficient test feedback:

{"tool": "audit_test", "kwargs": {"path": "/path/to/project", "mode": "failures"}}

Mode	~Tokens	Use case
`compact`	30	Fast-feedback loops (pass/fail + counts)
`failures`	30-200	First run, debugging (includes failure details)
`delta`	100-300	Post-refactor coverage comparison
`targeted`	50-150	Impact-scoped tests with `files` or `markers`

Output Format

The MCP tool returns the same structure as axm-audit audit . --agent:

Key	Type	Content
`score`	`float`	Composite quality score (0–100)
`grade`	`str`	Letter grade A–F
`passed`	`list`	Strings or dicts with actionable details
`failed`	`list`	Dicts with `rule_id`, `message`, `details`, `fix_hint`

For scoring details, see Scoring & Grades.