How ogiwbuz188/adaptive-fuzzer works
Adaptive Behavioral Fuzzer is a research prototype for reproducible, sandboxed fuzzing of 100 synthetic REST targets. Its main flow loads a benchmark manifest, derives target runs and seeds, launches targets and a fuzzer worker in Bubblewrap, discovers routes, mutates requests, analyzes responses, and records telemetry and reports. Optional branches provide authorization probes and a path-traversal canary; evaluation consumes recorded results and verified labels. The README documents several capabilities whose implementations are unsampled, so those nodes and links are limited to documented relationships.
Main entry points: sandbox_worker.py Sandbox worker · fuzzer.py Behavioral fuzzer · runner.py Benchmark runner · sandbox.py Bubblewrap sandbox
Components 17
Run execution
- Benchmark manifest
src/execution/manifest.py - Benchmark runner
src/execution/runner.py - Bubblewrap sandbox
src/execution/sandbox.py - Synthetic API targets
research_benchmark - Sandbox worker
src/execution/sandbox_worker.py
Fuzzing and detection
- Behavioral fuzzer
src/fuzzer.py - Statistical detector
src/detectors/statistical.py - Fingerprint detector
src/detectors/response_fingerprint.py - Path canary oracle
src/detectors/path_oracle.py - Authorization probe
src/detectors/bola.py
Telemetry and provenance
- Telemetry events
src/telemetry/schema.py - Telemetry writer
src/telemetry/writer.py - Run artifacts
Evaluation
- Evaluation metrics
src/evaluation/metrics.py - Statistical analysis
src/evaluation/statistical_analysis.py - Verified labels
src/ground_truth.py
Other components
- Researcher
Connections 21
- Researcher → Benchmark runner: starts runs
README.md - Benchmark runner → Benchmark manifest: loads
src/execution/runner.py - Benchmark runner → Bubblewrap sandbox: builds sandbox command
src/execution/runner.py - Bubblewrap sandbox → Synthetic API targets: isolates target
README.md - Bubblewrap sandbox → Sandbox worker: launches worker
README.md - Sandbox worker → Behavioral fuzzer: runs fuzz session
src/execution/sandbox_worker.py - Sandbox worker → Authorization probe: runs scenarios
src/execution/sandbox_worker.py - Sandbox worker → Benchmark manifest: reads benchmark settings
src/execution/sandbox_worker.py - Behavioral fuzzer → Synthetic API targets: sends requests
README.md - Behavioral fuzzer → Statistical detector: analyzes responses
src/fuzzer.py - Behavioral fuzzer → Fingerprint detector: checks fingerprints
src/fuzzer.py - Behavioral fuzzer → Path canary oracle: checks canary token
README.md - Sandbox worker → Telemetry events: creates events
src/execution/sandbox_worker.py - Sandbox worker → Telemetry writer: appends events
src/execution/sandbox_worker.py - Telemetry writer → Telemetry events: serializes events
src/telemetry/writer.py - Telemetry writer → Run artifacts: writes telemetry
README.md - Benchmark runner → Run artifacts: writes logs and reports
src/execution/runner.py - Researcher → Evaluation metrics: evaluates run
README.md - Evaluation metrics → Run artifacts: reads telemetry
README.md - Evaluation metrics → Verified labels: checks labels
README.md - Researcher → Statistical analysis: compares detector modes
README.md