SARIF Output Reference

GitHub, VS Code, and every major SAST dashboard speaks SARIF. Firmis outputs SARIF 2.1.0 natively — so your findings land directly in the GitHub Security tab, the VS Code Problems panel, and any CI security dashboard your team uses.

What is SARIF?

SARIF (Static Analysis Results Interchange Format) is an OASIS open standard (OASIS SARIF 2.1.0) for representing static analysis results as structured JSON. It defines a common schema so that any tool can produce results that any viewer can consume.

Key benefits of SARIF output:

Benefit	Description
GitHub Security tab	Upload via github/codeql-action/upload-sarif — findings appear as code scanning alerts
PR annotations	GitHub annotates pull request diffs with finding locations and messages
VS Code viewer	The SARIF Viewer extension (Microsoft) renders findings in the Problems panel
CI dashboards	Tools like Semgrep App, SonarQube, and Snyk Code accept SARIF imports
Historical tracking	Compare SARIF files across runs to track remediation progress

---

Generating SARIF output

Terminal

# Print SARIF to stdout
npx firmis scan --sarif

# Save SARIF to a file
npx firmis scan --sarif --output results.sarif

# Scan only MCP servers, output SARIF
npx firmis scan --platform mcp --sarif --output mcp-results.sarif

# Full CI pipeline with SARIF output
npx firmis ci --fail-on high --format sarif --output results.sarif

---

Field mapping

How Firmis finding fields map to SARIF 2.1.0 fields:

Firmis Field	SARIF Field	Notes
threat.ruleId	result.ruleId	e.g., tp-001, sd-045
threat.message	result.message.text	Human-readable finding description
threat.severity	result.level	See severity mapping table below
threat.location.file	result.locations[].physicalLocation.artifactLocation.uri	Relative path from scan root
threat.location.line	result.locations[].physicalLocation.region.startLine	1-indexed line number
threat.location.column	result.locations[].physicalLocation.region.startColumn	1-indexed column number
threat.evidence[].snippet	result.locations[].physicalLocation.region.snippet.text	Code snippet at finding location
threat.remediation	result.fixes[] or result.message	Remediation guidance when available
rule.name	run.tool.driver.rules[].name	Human-readable rule name
rule.description	run.tool.driver.rules[].fullDescription.text	Full rule description
rule.severity	run.tool.driver.rules[].defaultConfiguration.level	Rule’s default severity level

Severity mapping

SARIF uses a different severity vocabulary than Firmis:

Firmis Severity	SARIF Level
critical	error
high	error
medium	warning
low	note

Note

Both critical and high map to SARIF error level. This is standard practice — SARIF only has three actionable levels (error, warning, note). Use Firmis’s --fail-on critical or --fail-on high to distinguish between the two in CI logic.

---

Example SARIF document

results.sarif

{
  "$schema": "https://raw.githubusercontent.com/oasis-tcs/sarif-spec/master/Schemata/sarif-schema-2.1.0.json",
  "version": "2.1.0",
  "runs": [
    {
      "tool": {
        "driver": {
          "name": "Firmis",
          "version": "1.3.0",
          "informationUri": "https://firmislabs.com",
          "rules": [
            {
              "id": "tp-001",
              "name": "HiddenInstructionsInToolDescriptions",
              "shortDescription": {
                "text": "Hidden Instructions in Tool Descriptions"
              },
              "fullDescription": {
                "text": "Detects invisible Unicode characters and HTML comments used to hide malicious instructions inside tool or function descriptions."
              },
              "defaultConfiguration": {
                "level": "error"
              },
              "helpUri": "https://docs.firmislabs.com/reference/threat-categories#tool-poisoning"
            },
            {
              "id": "sd-045",
              "name": "OpenAIApiKeyDetected",
              "shortDescription": {
                "text": "OpenAI API Key Detected"
              },
              "fullDescription": {
                "text": "A hardcoded OpenAI API key was found in the source file."
              },
              "defaultConfiguration": {
                "level": "error"
              },
              "helpUri": "https://docs.firmislabs.com/reference/threat-categories#secret-detection"
            }
          ]
        }
      },
      "results": [
        {
          "ruleId": "tp-001",
          "level": "error",
          "message": {
            "text": "Hidden instructions in tool description: zero-width space (U+200B) detected."
          },
          "locations": [
            {
              "physicalLocation": {
                "artifactLocation": {
                  "uri": "src/tools/search.ts",
                  "uriBaseId": "%SRCROOT%"
                },
                "region": {
                  "startLine": 14,
                  "startColumn": 18,
                  "snippet": {
                    "text": "description: \"Search the web\u200B and return results\""
                  }
                }
              }
            }
          ]
        },
        {
          "ruleId": "sd-045",
          "level": "error",
          "message": {
            "text": "Hardcoded OpenAI API key detected. Rotate this key immediately and store it in an environment variable."
          },
          "locations": [
            {
              "physicalLocation": {
                "artifactLocation": {
                  "uri": "config/llm.ts",
                  "uriBaseId": "%SRCROOT%"
                },
                "region": {
                  "startLine": 12,
                  "startColumn": 15
                }
              }
            }
          ]
        }
      ]
    }
  ]
}

---

Uploading to GitHub Security tab

GitHub’s code scanning feature accepts SARIF files uploaded via the upload-sarif action.

.github/workflows/firmis.yml

name: Firmis Security Scan

on:
  push:
    branches: [main]
  pull_request:
    branches: [main]

jobs:
  scan:
    runs-on: ubuntu-latest
    permissions:
      security-events: write
      contents: read

    steps:
      - uses: actions/checkout@v4

      - name: Run Firmis scan
        run: npx firmis scan --sarif --output results.sarif
        continue-on-error: true

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

Tip

Set continue-on-error: true on the scan step so the SARIF upload always runs, even when findings are present. The upload-sarif action will show the findings in the Security tab regardless of whether the scan step passed or failed.

After uploading, findings appear at Security → Code scanning alerts in your GitHub repository. Pull request diffs are annotated with inline finding markers at the relevant line numbers.

---

Viewing SARIF locally

VS Code SARIF Viewer

Install the SARIF Viewer extension from Microsoft. Open any .sarif file in VS Code to browse findings in the Problems panel with source location highlighting.

SARIF Web Viewer

Microsoft maintains a browser-based viewer at microsoft.github.io/sarif-web-component. Upload your .sarif file to inspect results without installing anything.

---

SARIF in the CI pipeline

The firmis ci command generates SARIF as part of the full discover → BOM → scan → report pipeline:

Terminal

npx firmis ci --fail-on high --format sarif --output results.sarif

This produces results.sarif alongside agent-bom.json. Both can be archived as CI artifacts:

.github/workflows/firmis-ci.yml (excerpt)

- name: Run Firmis CI pipeline
  run: npx firmis ci --fail-on high --format sarif --output results.sarif

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

- name: Archive Agent BOM
  uses: actions/upload-artifact@v4
  with:
    name: agent-bom
    path: agent-bom.json

---

What to do next

• firmis scan → — CLI reference including --sarif flag
• firmis ci → — CI pipeline command
• GitHub Actions integration → — full workflow example with SARIF upload
• CycloneDX BOM → — the agent inventory output format
• Threat Categories → — what each ruleId maps to across 16 categories