Output Formats

Aguara supports four output formats optimized for different use cases: human-readable terminal output with ANSI colors, machine-parseable JSON, SARIF for GitHub Code Scanning, and Markdown for PR comments and job summaries.

Terminal (Default)

Use Case: Interactive development, local scans, debugging

aguara scan ./skills/

Example output:

────────────────────────────────────────────────────────────────────────
  AGUARA SCAN RESULTS
  Target: ./skills/  ·  12 files  ·  177 rules  ·  0.84s
────────────────────────────────────────────────────────────────────────

  CRITICAL   ████████████████████████████████        12
  HIGH       ████████████████                         6
  MEDIUM     ████████                                 3

  21 findings

── CRITICAL (12) ──────────────────────────────────────────────────────

  skills/deploy.md

    ✖ PROMPT_INJECTION_001      Instruction override attempt            skills/deploy.md:15
      │ Ignore all previous instructions

    ✖ CRED_001                  OpenAI API key detected                 skills/deploy.md:23
      │ sk-proj-abc123...

── HIGH (6) ───────────────────────────────────────────────────────────

  skills/search.md

    ▲ EXFIL_005                 Webhook to external domain              skills/search.md:42
    ▲ CMD_EXEC_003              Shell command execution                 skills/search.md:48

── TOP AFFECTED FILES ────────────────────────────────────────────────

     8  skills/deploy.md
     5  skills/search.md
     3  skills/admin.md

────────────────────────────────────────────────────────────────────────
  12 files scanned · 21 findings · 177 rules · 0.84s
────────────────────────────────────────────────────────────────────────

Features

ANSI Colors: Severity-coded icons and bars (red=critical, yellow=medium, etc.)
Severity Dashboard: Horizontal bar chart of findings by severity
Grouped by File: Findings organized by file path
Top Files Chart: Most-affected files ranked by finding count
Preview Text: Matched text snippet for each finding
Code Block Tags: [code] marker for findings in fenced code blocks
Confidence Scores: [85%] shown with --verbose

Flags

# Disable colors (for non-TTY environments)
aguara scan --no-color .

# Verbose mode (show descriptions, confidence, remediation)
aguara scan --verbose .

Environment variable:

export NO_COLOR=1
aguara scan .

JSON

Use Case: Machine processing, custom tooling, API integration, dashboards

aguara scan --format json ./skills/

Example output:

{
  "findings": [
    {
      "rule_id": "PROMPT_INJECTION_001",
      "rule_name": "Instruction override attempt",
      "severity": 4,
      "category": "prompt-injection",
      "description": "Detects attempts to override or ignore previous instructions",
      "file_path": "skills/deploy.md",
      "line": 15,
      "column": 0,
      "matched_text": "Ignore all previous instructions",
      "context": [
        {"line": 14, "content": "", "is_match": false},
        {"line": 15, "content": "Ignore all previous instructions", "is_match": true},
        {"line": 16, "content": "", "is_match": false}
      ],
      "score": 60.0,
      "analyzer": "pattern",
      "in_code_block": false,
      "confidence": 0.85,
      "remediation": "Remove instruction override text. If this is documentation, wrap it in a code block."
    }
  ],
  "files_scanned": 12,
  "rules_loaded": 177,
  "duration_ms": 840
}

Schema

Root Object

Field	Type	Description
`findings`	`Finding[]`	Array of all findings
`files_scanned`	`int`	Number of files analyzed
`rules_loaded`	`int`	Number of active detection rules
`duration_ms`	`int`	Scan duration in milliseconds

Finding Object

Field	Type	Description
`rule_id`	`string`	Rule identifier (e.g., `PROMPT_INJECTION_001`)
`rule_name`	`string`	Human-readable rule name
`severity`	`int`	0=INFO, 1=LOW, 2=MEDIUM, 3=HIGH, 4=CRITICAL
`category`	`string`	Rule category (e.g., `prompt-injection`)
`description`	`string`	Rule description
`file_path`	`string`	Relative path to file
`line`	`int`	Line number (1-indexed)
`column`	`int`	Column number (0-indexed)
`matched_text`	`string`	Text that triggered the rule
`context`	`ContextLine[]`	Surrounding lines
`score`	`float`	Risk score (0-100)
`analyzer`	`string`	`pattern`, `nlp-injection`, `toxicflow`, or `rugpull`
`in_code_block`	`bool`	True if inside markdown code fence
`confidence`	`float`	Confidence score (0.0-1.0)
`remediation`	`string`	Fix guidance (optional)

ContextLine Object

Field	Type	Description
`line`	`int`	Line number
`content`	`string`	Line text
`is_match`	`bool`	True if this is the matched line

Save to File

aguara scan --format json -o results.json ./skills/

SARIF

Use Case: GitHub Code Scanning, IDE integrations, SAST dashboards

aguara scan --format sarif ./skills/

Example output:

{
  "$schema": "https://docs.oasis-open.org/sarif/sarif/v2.1.0/sarif-schema-2.1.0.json",
  "version": "2.1.0",
  "runs": [
    {
      "tool": {
        "driver": {
          "name": "aguara",
          "version": "v0.5.0",
          "informationUri": "https://github.com/garagon/aguara",
          "rules": [
            {
              "id": "PROMPT_INJECTION_001",
              "name": "Instruction override attempt",
              "shortDescription": {"text": "Instruction override attempt"},
              "defaultConfiguration": {"level": "error"},
              "properties": {"tags": ["prompt-injection"]}
            }
          ]
        }
      },
      "results": [
        {
          "ruleId": "PROMPT_INJECTION_001",
          "ruleIndex": 0,
          "level": "error",
          "message": {"text": "Instruction override attempt: Ignore all previous instructions"},
          "locations": [
            {
              "physicalLocation": {
                "artifactLocation": {"uri": "skills/deploy.md"},
                "region": {"startLine": 15, "startColumn": 1}
              }
            }
          ],
          "properties": {
            "confidence": 0.85,
            "rank": 85.0
          }
        }
      ],
      "properties": {"duration_ms": 840}
    }
  ]
}

SARIF 2.1.0 Compliance

Aguara outputs valid SARIF 2.1.0 for:

GitHub Code Scanning: Upload with actions/upload-sarif
GitLab SAST: Store as artifact in sast report type
Visual Studio: Load into Error List
VS Code: Render with SARIF Viewer extension

Severity Mapping

Aguara Severity	SARIF Level
CRITICAL	`error`
HIGH	`warning`
MEDIUM	`note`
LOW	`note`
INFO	`none`

GitHub Code Scanning

Upload SARIF results to GitHub Security tab:

# .github/workflows/security.yml
name: Security Scan

on:
  push:
    branches: [main]
  pull_request:

jobs:
  scan:
    runs-on: ubuntu-latest
    permissions:
      security-events: write  # required for SARIF upload
    steps:
      - uses: actions/checkout@v4

      - name: Run Aguara
        run: |
          curl -fsSL https://raw.githubusercontent.com/garagon/aguara/main/install.sh | bash
          aguara scan --format sarif -o results.sarif .

      - name: Upload SARIF
        uses: github/codeql-action/upload-sarif@v3
        with:
          sarif_file: results.sarif

Findings appear in the Security tab → Code scanning alerts.

Markdown

Use Case: GitHub Actions job summaries, PR comments, documentation

aguara scan --format markdown ./skills/

Example output:

## Aguara Security Scan

> :red_circle: **12 critical** &nbsp; :orange_circle: **6 high** &nbsp; :yellow_circle: **3 medium** · `./skills/` · 12 files · 177 rules · 0.84s

### `skills/deploy.md`

| Severity | Rule | Finding | Line |
|:--------:|------|---------|-----:|
| :red_circle: | `PROMPT_INJECTION_001` | Instruction override attempt | 15 |
| :red_circle: | `CRED_001` | OpenAI API key detected | 23 |

### `skills/search.md`

| Severity | Rule | Finding | Line |
|:--------:|------|---------|-----:|
| :orange_circle: | `EXFIL_005` | Webhook to external domain | 42 |
| :orange_circle: | `CMD_EXEC_003` | Shell command execution | 48 |

---

<sub>Generated by [Aguara](https://github.com/garagon/aguara) v0.5.0 — security scanner for AI agent skills and MCP servers</sub>

GitHub Actions Job Summary

Render Markdown output in the Actions summary:

- name: Scan and summarize
  run: |
    aguara scan --format markdown . >> $GITHUB_STEP_SUMMARY

Appears in the Actions run summary page.

PR Comments

Post scan results as a PR comment:

- name: Scan
  run: aguara scan --format markdown . > scan-results.md

- name: Comment PR
  uses: actions/github-script@v7
  with:
    script: |
      const fs = require('fs');
      const body = fs.readFileSync('scan-results.md', 'utf8');
      github.rest.issues.createComment({
        issue_number: context.issue.number,
        owner: context.repo.owner,
        repo: context.repo.repo,
        body: body
      });

Format Comparison

Format	Best For	Size	Structured	Human-Readable
Terminal	Local dev, debugging	N/A	No	✅
JSON	APIs, dashboards, custom tools	Medium	✅	No
SARIF	GitHub Code Scanning, IDEs	Large	✅	No
Markdown	PR comments, job summaries	Small	Semi	✅

Writing to Files

All formats support -o / --output:

# JSON to file
aguara scan --format json -o results.json .

# SARIF to file
aguara scan --format sarif -o results.sarif .

# Markdown to file
aguara scan --format markdown -o summary.md .

# Terminal to file (preserves colors if TTY detected)
aguara scan -o scan-log.txt .

Default is stdout.

Filtering Output

Combine --severity to reduce noise:

# Only critical and high
aguara scan --format json --severity high . > critical.json

# Only medium and above
aguara scan --format sarif --severity medium -o findings.sarif .

Exit Codes with —fail-on

Control exit code based on severity:

# Exit 1 if high or critical findings
aguara scan --format json --fail-on high . > results.json
echo $?  # 1 if findings, 0 otherwise

Useful for CI pipelines that need to fail on security issues.

Incremental Scanning — Scan only changed files
Rug-Pull Detection — Track changes across scans
MCP Discovery — Auto-detect and scan MCP configs

Get Started

Core Concepts

CLI Usage

Configuration

CI/CD Integration

Rules & Detection

Advanced Features

Output Formats

Terminal (Default)

Features

Flags

JSON

Schema

Root Object

Finding Object

ContextLine Object

Save to File

SARIF

SARIF 2.1.0 Compliance

Severity Mapping

GitHub Code Scanning

Markdown

GitHub Actions Job Summary

PR Comments

Format Comparison

Writing to Files

Filtering Output

Exit Codes with —fail-on

Build docs developers (and LLMs) love

Get Started

Core Concepts

CLI Usage

Configuration

CI/CD Integration

Rules & Detection

Advanced Features

Documentation Index

​Terminal (Default)

​Features

​Flags

​JSON

​Schema

​Root Object

​Finding Object

​ContextLine Object

​Save to File

​SARIF

​SARIF 2.1.0 Compliance

​Severity Mapping

​GitHub Code Scanning

​Markdown

​GitHub Actions Job Summary

​PR Comments

​Format Comparison

​Writing to Files

​Filtering Output

​Exit Codes with —fail-on

​Related

Build docs developers (and LLMs) love

Terminal (Default)

Features

Flags

JSON

Schema

Root Object

Finding Object

ContextLine Object

Save to File

SARIF

SARIF 2.1.0 Compliance

Severity Mapping

GitHub Code Scanning

Markdown

GitHub Actions Job Summary

PR Comments

Format Comparison

Writing to Files

Filtering Output

Exit Codes with —fail-on

Related