CalvinBackup/fuzzforge_ai
1
0
Fork 0
mirror of https://github.com/FuzzingLabs/fuzzforge_ai.git synced 2026-02-13 01:12:45 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
60ca088ecfa439c9d2ddd27f694ee8397dbd3f54
fuzzforge_ai/ai
History
tduhamel42 60ca088ecf CI/CD Integration with Ephemeral Deployment Model (#14) 
* feat: Complete migration from Prefect to Temporal

BREAKING CHANGE: Replaces Prefect workflow orchestration with Temporal

## Major Changes
- Replace Prefect with Temporal for workflow orchestration
- Implement vertical worker architecture (rust, android)
- Replace Docker registry with MinIO for unified storage
- Refactor activities to be co-located with workflows
- Update all API endpoints for Temporal compatibility

## Infrastructure
- New: docker-compose.temporal.yaml (Temporal + MinIO + workers)
- New: workers/ directory with rust and android vertical workers
- New: backend/src/temporal/ (manager, discovery)
- New: backend/src/storage/ (S3-cached storage with MinIO)
- New: backend/toolbox/common/ (shared storage activities)
- Deleted: docker-compose.yaml (old Prefect setup)
- Deleted: backend/src/core/prefect_manager.py
- Deleted: backend/src/services/prefect_stats_monitor.py
- Deleted: Docker registry and insecure-registries requirement

## Workflows
- Migrated: security_assessment workflow to Temporal
- New: rust_test workflow (example/test workflow)
- Deleted: secret_detection_scan (Prefect-based, to be reimplemented)
- Activities now co-located with workflows for independent testing

## API Changes
- Updated: backend/src/api/workflows.py (Temporal submission)
- Updated: backend/src/api/runs.py (Temporal status/results)
- Updated: backend/src/main.py (727 lines, TemporalManager integration)
- Updated: All 16 MCP tools to use TemporalManager

## Testing
- ✅ All services healthy (Temporal, PostgreSQL, MinIO, workers, backend)
- ✅ All API endpoints functional
- ✅ End-to-end workflow test passed (72 findings from vulnerable_app)
- ✅ MinIO storage integration working (target upload/download, results)
- ✅ Worker activity discovery working (6 activities registered)
- ✅ Tarball extraction working
- ✅ SARIF report generation working

## Documentation
- ARCHITECTURE.md: Complete Temporal architecture documentation
- QUICKSTART_TEMPORAL.md: Getting started guide
- MIGRATION_DECISION.md: Why we chose Temporal over Prefect
- IMPLEMENTATION_STATUS.md: Migration progress tracking
- workers/README.md: Worker development guide

## Dependencies
- Added: temporalio>=1.6.0
- Added: boto3>=1.34.0 (MinIO S3 client)
- Removed: prefect>=3.4.18

* feat: Add Python fuzzing vertical with Atheris integration

This commit implements a complete Python fuzzing workflow using Atheris:

## Python Worker (workers/python/)
- Dockerfile with Python 3.11, Atheris, and build tools
- Generic worker.py for dynamic workflow discovery
- requirements.txt with temporalio, boto3, atheris dependencies
- Added to docker-compose.temporal.yaml with dedicated cache volume

## AtherisFuzzer Module (backend/toolbox/modules/fuzzer/)
- Reusable module extending BaseModule
- Auto-discovers fuzz targets (fuzz_*.py, *_fuzz.py, fuzz_target.py)
- Recursive search to find targets in nested directories
- Dynamically loads TestOneInput() function
- Configurable max_iterations and timeout
- Real-time stats callback support for live monitoring
- Returns findings as ModuleFinding objects

## Atheris Fuzzing Workflow (backend/toolbox/workflows/atheris_fuzzing/)
- Temporal workflow for orchestrating fuzzing
- Downloads user code from MinIO
- Executes AtherisFuzzer module
- Uploads results to MinIO
- Cleans up cache after execution
- metadata.yaml with vertical: python for routing

## Test Project (test_projects/python_fuzz_waterfall/)
- Demonstrates stateful waterfall vulnerability
- main.py with check_secret() that leaks progress
- fuzz_target.py with Atheris TestOneInput() harness
- Complete README with usage instructions

## Backend Fixes
- Fixed parameter merging in REST API endpoints (workflows.py)
- Changed workflow parameter passing from positional args to kwargs (manager.py)
- Default parameters now properly merged with user parameters

## Testing
✅ Worker discovered AtherisFuzzingWorkflow
✅ Workflow executed end-to-end successfully
✅ Fuzz target auto-discovered in nested directories
✅ Atheris ran 100,000 iterations
✅ Results uploaded and cache cleaned

* chore: Complete Temporal migration with updated CLI/SDK/docs

This commit includes all remaining Temporal migration changes:

## CLI Updates (cli/)
- Updated workflow execution commands for Temporal
- Enhanced error handling and exceptions
- Updated dependencies in uv.lock

## SDK Updates (sdk/)
- Client methods updated for Temporal workflows
- Updated models for new workflow execution
- Updated dependencies in uv.lock

## Documentation Updates (docs/)
- Architecture documentation for Temporal
- Workflow concept documentation
- Resource management documentation (new)
- Debugging guide (new)
- Updated tutorials and how-to guides
- Troubleshooting updates

## README Updates
- Main README with Temporal instructions
- Backend README
- CLI README
- SDK README

## Other
- Updated IMPLEMENTATION_STATUS.md
- Removed old vulnerable_app.tar.gz

These changes complete the Temporal migration and ensure the
CLI/SDK work correctly with the new backend.

* fix: Use positional args instead of kwargs for Temporal workflows

The Temporal Python SDK's start_workflow() method doesn't accept
a 'kwargs' parameter. Workflows must receive parameters as positional
arguments via the 'args' parameter.

Changed from:
  args=workflow_args  # Positional arguments

This fixes the error:
  TypeError: Client.start_workflow() got an unexpected keyword argument 'kwargs'

Workflows now correctly receive parameters in order:
- security_assessment: [target_id, scanner_config, analyzer_config, reporter_config]
- atheris_fuzzing: [target_id, target_file, max_iterations, timeout_seconds]
- rust_test: [target_id, test_message]

* fix: Filter metadata-only parameters from workflow arguments

SecurityAssessmentWorkflow was receiving 7 arguments instead of 2-5.
The issue was that target_path and volume_mode from default_parameters
were being passed to the workflow, when they should only be used by
the system for configuration.

Now filters out metadata-only parameters (target_path, volume_mode)
before passing arguments to workflow execution.

* refactor: Remove Prefect leftovers and volume mounting legacy

Complete cleanup of Prefect migration artifacts:

Backend:
- Delete registry.py and workflow_discovery.py (Prefect-specific files)
- Remove Docker validation from setup.py (no longer needed)
- Remove ResourceLimits and VolumeMount models
- Remove target_path and volume_mode from WorkflowSubmission
- Remove supported_volume_modes from API and discovery
- Clean up metadata.yaml files (remove volume/path fields)
- Simplify parameter filtering in manager.py

SDK:
- Remove volume_mode parameter from client methods
- Remove ResourceLimits and VolumeMount models
- Remove Prefect error patterns from docker_logs.py
- Clean up WorkflowSubmission and WorkflowMetadata models

CLI:
- Remove Volume Modes display from workflow info

All removed features are Prefect-specific or Docker volume mounting
artifacts. Temporal workflows use MinIO storage exclusively.

* feat: Add comprehensive test suite and benchmark infrastructure

- Add 68 unit tests for fuzzer, scanner, and analyzer modules
- Implement pytest-based test infrastructure with fixtures
- Add 6 performance benchmarks with category-specific thresholds
- Configure GitHub Actions for automated testing and benchmarking
- Add test and benchmark documentation

Test coverage:
- AtherisFuzzer: 8 tests
- CargoFuzzer: 14 tests
- FileScanner: 22 tests
- SecurityAnalyzer: 24 tests

All tests passing (68/68)
All benchmarks passing (6/6)

* fix: Resolve all ruff linting violations across codebase

Fixed 27 ruff violations in 12 files:
- Removed unused imports (Depends, Dict, Any, Optional, etc.)
- Fixed undefined workflow_info variable in workflows.py
- Removed dead code with undefined variables in atheris_fuzzer.py
- Changed f-string to regular string where no placeholders used

All files now pass ruff checks for CI/CD compliance.

* fix: Configure CI for unit tests only

- Renamed docker-compose.temporal.yaml → docker-compose.yml for CI compatibility
- Commented out integration-tests job (no integration tests yet)
- Updated test-summary to only depend on lint and unit-tests

CI will now run successfully with 68 unit tests. Integration tests can be added later.

* feat: Add CI/CD integration with ephemeral deployment model

Implements comprehensive CI/CD support for FuzzForge with on-demand worker management:

**Worker Management (v0.7.0)**
- Add WorkerManager for automatic worker lifecycle control
- Auto-start workers from stopped state when workflows execute
- Auto-stop workers after workflow completion
- Health checks and startup timeout handling (90s default)

**CI/CD Features**
- `--fail-on` flag: Fail builds based on SARIF severity levels (error/warning/note/info)
- `--export-sarif` flag: Export findings in SARIF 2.1.0 format
- `--auto-start`/`--auto-stop` flags: Control worker lifecycle
- Exit code propagation: Returns 1 on blocking findings, 0 on success

**Exit Code Fix**
- Add `except typer.Exit: raise` handlers at 3 critical locations
- Move worker cleanup to finally block for guaranteed execution
- Exit codes now propagate correctly even when build fails

**CI Scripts & Examples**
- ci-start.sh: Start FuzzForge services with health checks
- ci-stop.sh: Clean shutdown with volume preservation option
- GitHub Actions workflow example (security-scan.yml)
- GitLab CI pipeline example (.gitlab-ci.example.yml)
- docker-compose.ci.yml: CI-optimized compose file with profiles

**OSS-Fuzz Integration**
- New ossfuzz_campaign workflow for running OSS-Fuzz projects
- OSS-Fuzz worker with Docker-in-Docker support
- Configurable campaign duration and project selection

**Documentation**
- Comprehensive CI/CD integration guide (docs/how-to/cicd-integration.md)
- Updated architecture docs with worker lifecycle details
- Updated workspace isolation documentation
- CLI README with worker management examples

**SDK Enhancements**
- Add get_workflow_worker_info() endpoint
- Worker vertical metadata in workflow responses

**Testing**
- All workflows tested: security_assessment, atheris_fuzzing, secret_detection, cargo_fuzzing
- All monitoring commands tested: stats, crashes, status, finding
- Full CI pipeline simulation verified
- Exit codes verified for success/failure scenarios

Ephemeral CI/CD model: ~3-4GB RAM, ~60-90s startup, runs entirely in CI containers.

* fix: Resolve ruff linting violations in CI/CD code

- Remove unused variables (run_id, defaults, result)
- Remove unused imports
- Fix f-string without placeholders

All CI/CD integration files now pass ruff checks.
2025-10-14 10:13:45 +02:00
..
src/fuzzforge_ai
CI/CD Integration with Ephemeral Deployment Model (#14)
2025-10-14 10:13:45 +02:00
.gitignore
Initial commit
2025-09-29 21:26:41 +02:00
llm.txt
Initial commit
2025-09-29 21:26:41 +02:00
pyproject.toml
Initial commit
2025-09-29 21:26:41 +02:00
README.md
Initial commit
2025-09-29 21:26:41 +02:00

README.md

FuzzForge AI Module

FuzzForge AI is the multi-agent layer that lets you operate the FuzzForge security platform through natural language. It orchestrates local tooling, registered Agent-to-Agent (A2A) peers, and the Prefect-powered backend while keeping long-running context in memory and project knowledge graphs.

Quick Start

  1. Initialise a project 
    cd /path/to/project
fuzzforge init
  2. Review environment settings – copy .fuzzforge/.env.template to .fuzzforge/.env, then edit the values to match your provider. The template ships with commented defaults for OpenAI-style usage and placeholders for Cognee keys. 
    LLM_PROVIDER=openai
LITELLM_MODEL=gpt-5-mini
OPENAI_API_KEY=sk-your-key
FUZZFORGE_MCP_URL=http://localhost:8010/mcp
SESSION_PERSISTENCE=sqlite
    Optional flags you may want to enable early: 
    MEMORY_SERVICE=inmemory
AGENTOPS_API_KEY=sk-your-agentops-key   # Enable hosted tracing
LOG_LEVEL=INFO                          # CLI / server log level
  3. Populate the knowledge graph 
    fuzzforge ingest --path . --recursive
# alias: fuzzforge rag ingest --path . --recursive
  4. Launch the agent shell 
    fuzzforge ai agent
    Keep the backend running (Prefect API at FUZZFORGE_MCP_URL) so workflow commands succeed.

Everyday Workflow

  • Run fuzzforge ai agent and start with list available fuzzforge workflows or /memory status to confirm everything is wired.
  • Use natural prompts for automation (run fuzzforge workflow …, search project knowledge for …) and fall back to slash commands for precision (/recall, /sendfile).
  • Keep /memory datasets handy to see which Cognee datasets are available after each ingest.
  • Start the HTTP surface with python -m fuzzforge_ai when external agents need access to artifacts or graph queries. The CLI stays usable at the same time.
  • Refresh the knowledge graph regularly: fuzzforge ingest --path . --recursive --force keeps responses aligned with recent code changes.

What the Agent Can Do

  • Route requests – automatically selects the right local tool or remote agent using the A2A capability registry.
  • Run security workflows – list, submit, and monitor FuzzForge workflows via MCP wrappers.
  • Manage artifacts – create downloadable files for reports, code edits, and shared attachments.
  • Maintain context – stores session history, semantic recall, and Cognee project graphs.
  • Serve over HTTP – expose the same agent as an A2A server using python -m fuzzforge_ai.

Essential Commands

Inside fuzzforge ai agent you can mix slash commands and free-form prompts:

/list                     # Show registered A2A agents
/register http://:10201   # Add a remote agent
/artifacts                 # List generated files
/sendfile SecurityAgent src/report.md "Please review"
You> route_to SecurityAnalyzer: scan ./backend for secrets
You> run fuzzforge workflow static_analysis_scan on ./test_projects/demo
You> search project knowledge for "prefect status" using INSIGHTS

Artifacts created during the conversation are served from .fuzzforge/artifacts/ and exposed through the A2A HTTP API.

Memory & Knowledge

The module layers three storage systems:

  • Session persistence (SQLite or in-memory) for chat transcripts.
  • Semantic recall via the ADK memory service for fuzzy search.
  • Cognee graphs for project-wide knowledge built from ingestion runs.

Re-run ingestion after major code changes to keep graph answers relevant. If Cognee variables are not set, graph-specific tools automatically respond with a polite "not configured" message.

Sample Prompts

Use these to validate the setup once the agent shell is running:

  • list available fuzzforge workflows
  • run fuzzforge workflow static_analysis_scan on ./backend with target_branch=main
  • show findings for that run once it finishes
  • refresh the project knowledge graph for ./backend
  • search project knowledge for "prefect readiness" using INSIGHTS
  • /recall terraform secrets
  • /memory status
  • ROUTE_TO SecurityAnalyzer: audit infrastructure_vulnerable

Need More Detail?

Dive into the dedicated guides under ai/docs/advanced/:

  • Architecture – High-level architecture with diagrams and component breakdowns.
  • Ingestion – Command options, Cognee persistence, and prompt examples.
  • Configuration – LLM provider matrix, local model setup, and tracing options.
  • Prompts – Slash commands, workflow prompts, and routing tips.
  • A2A Services – HTTP endpoints, agent card, and collaboration flow.
  • Memory Persistence – Deep dive on memory storage, datasets, and how /memory status inspects them.

Development Notes

  • Entry point for the CLI: ai/src/fuzzforge_ai/cli.py
  • A2A HTTP server: ai/src/fuzzforge_ai/a2a_server.py
  • Tool routing & workflow glue: ai/src/fuzzforge_ai/agent_executor.py
  • Ingestion helpers: ai/src/fuzzforge_ai/ingest_utils.py

Install the module in editable mode (pip install -e ai) while iterating so CLI changes are picked up immediately.

Reference in New Issue View Git Blame Copy Permalink