Usage Modes¶
MethodAtlas operates in three independent modes. A source write-back modifier can be layered on top of any mode that produces AI output.
| Mode | AI required | Network | Output formats |
|---|---|---|---|
| Static inventory | No | No | CSV, plain text, SARIF |
| API AI enrichment | Yes — via API | Yes | CSV, plain text, SARIF |
| Manual AI workflow | Yes — via chat | No | CSV, plain text, SARIF |
| Source write-back | Combined with above | Depends | — |
Choose the mode that fits your infrastructure constraints:
- No AI access at all → Static inventory gives you the method list; tags can be added manually or in a later enrichment step.
- Direct API access → API AI enrichment is the simplest path; all four hosted providers (Ollama, OpenAI, OpenRouter, Anthropic) are supported.
- Air-gapped or policy-restricted → Manual AI workflow lets you use a supervised chat interface without any outbound network calls from the scan host.
YAML configuration¶
All modes support a shared configuration file that avoids repeating flags:
outputMode: csv # csv | plain | sarif
contentHash: false
ai:
provider: openrouter
model: google/gemini-flash-1.5
apiKeyEnv: OPENROUTER_API_KEY
confidence: false
CLI flags always override YAML values. See the CLI Reference for the full flag list and the complete YAML schema.