How mandiant/capa works
capa is a malware-capability analyzer: the CLI accepts executable samples or sandbox reports, selects an extractor, evaluates extracted features against rules, and presents capability results. Static analysis has multiple backend integrations; dynamic analysis consumes reports. Results can also be frozen or explored interactively in the browser, and IDA/Ghidra plugins provide analyst-facing integrations. Some subsystem links are documented only at a high level; the graph avoids asserting unshown calls.
Main entry points: main.py CLI orchestration · extractors Static extractors · base_extractor.py Extractor interface · loader.py Input loader
Components 24
Analysis entry
- CLI orchestration
capa/main.py - Input loader
capa/loader.py
Feature extraction
- Static extractors
capa/features/extractors - Sandbox extractors
capa/features/extractors - Extractor interface
capa/features/extractors/base_extractor.py - ELF analysis
capa/features/extractors/elf.py - Ghidra backend
capa/features/extractors/ghidra/extractor.py - Feature model
capa/features/common.py - Feature addresses
capa/features/address.py
Capability matching
- Rule loading
capa/rules/__init__.py - Rule evaluator
capa/engine.py - Capability analysis
capa/capabilities/common.py - Static matching
capa/capabilities/static.py - Dynamic matching
capa/capabilities/dynamic.py
Results and export
- Result document
capa/render/result_document.py - CLI renderers
capa/render - Feature freezing
capa/features/freeze
Analyst interfaces
- Browser Explorer
web/explorer/src/App.vue - Result loading
web/explorer/src/composables/useRdocLoader.js - Result parser
web/explorer/src/utils/rdocParser.js - IDA plugin
capa/ida/plugin/form.py - Ghidra plugin
capa/ghidra/plugin/capa_explorer.py
Other components
- Analyst
- Sample or report
Connections 28
- Analyst → CLI orchestration: runs analysis
- Sample or report → CLI orchestration: provides input
- CLI orchestration → Input loader: loads sample
capa/main.py - Input loader → Static extractors: selects backend
capa/loader.py - Input loader → Sandbox extractors: selects backend
capa/loader.py - Static extractors → Extractor interface: implements interface
capa/features/extractors/ghidra/extractor.py - Sandbox extractors → Extractor interface: implements interface
capa/features/extractors/ghidra/extractor.py - Static extractors → ELF analysis: includes backend
- Static extractors → Ghidra backend: includes backend
capa/features/extractors/ghidra/extractor.py - Ghidra backend → Extractor interface: uses interface
capa/features/extractors/ghidra/extractor.py - Extractor interface → Feature model: yields features
capa/features/extractors/base_extractor.py - Feature model → Feature addresses: uses addresses
capa/features/common.py - CLI orchestration → Rule loading: loads rules
capa/main.py - CLI orchestration → Capability analysis: analyzes features
capa/main.py - Capability analysis → Static matching: dispatches static
capa/capabilities/common.py - Capability analysis → Dynamic matching: dispatches dynamic
capa/capabilities/common.py - Static matching → Rule evaluator: evaluates rules
capa/capabilities/static.py - Dynamic matching → Rule evaluator: evaluates rules
capa/capabilities/dynamic.py - Static matching → Result document: records matches
capa/capabilities/static.py - Dynamic matching → Result document: records matches
capa/capabilities/dynamic.py - CLI orchestration → CLI renderers: renders results
capa/main.py - CLI renderers → Result document: formats document
capa/render/json.py - CLI orchestration → Feature freezing: can freeze features
capa/main.py - Analyst → Browser Explorer: explores results
- Browser Explorer → Result loading: loads documents
- Result loading → Result parser: parses results
- Analyst → IDA plugin: uses analysis plugin
- Analyst → Ghidra plugin: uses analysis plugin