rsolv-dev/RSOLV: Test-First AI Security Fixes

AI security that proves vulnerabilities with executable tests before fixing. Reduces false positives.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
api_keyRSOLV API key for authentication (deprecated, use rsolvApiKey)no
rsolvApiKeyRSOLV API key for authentication (get yours at rsolv.dev/signup)yes
config_pathPath to RSOLV configuration fileno.github/rsolv.yml
issue_labelLabel to identify issues for automationnorsolv:detected
environment_variablesJSON string of environment variables to pass to the containerno
api_urlRSOLV API URL (defaults to production)nohttps://api.rsolv.dev
github-tokenGitHub token for PR/issue operations. Automatically provided by GitHub Actions. Only override if you need custom permissions. no${{ github.token }}
modeOperation mode - choose based on your workflow: - 'scan': Scan repository for vulnerabilities, create GitHub issues, register a PipelineRun (recommended starting point) - 'process': Process pending work for an existing PipelineRun (use in a separate job after scan, pass pipeline_run_id) - 'full': Run all phases in a single job — scan, validate, and mitigate (simplest setup, uses more credits at once) - 'validate': Validate a single issue via backend orchestration (advanced, requires issue_number) - 'mitigate': Mitigate a single validated issue via backend orchestration (advanced, requires issue_number) Recommended patterns: - Single-job: mode=full (easiest, good for small repos) - Multi-job: Job 1: mode=scan → Job 2: mode=process with pipeline_run_id from Job 1 noscan
scan_modeScan mode operation: fix (default) or scannofix
enable_security_analysisEnable security vulnerability analysisnotrue
scan_outputWhere scan findings go (comma-separated). Options: issues, report, dashboard. - 'issues': Create GitHub issues (default, triggers validate/mitigate pipeline) - 'report': Upload structured JSON+markdown report as workflow artifact - 'dashboard': Send findings to RSOLV platform dashboard (future) Use 'report' alone for audit/discovery scans without issue creation. noissues
max_issuesMaximum number of issues to process in a single runno3
max_validationsMaximum validations to process per run. Capped by your remaining budget.no
issue_numberSpecific issue number to process (for workflow_dispatch)no""
use_git_based_editingEnable git-based in-place editing for true file modifications (ADR-012)notrue
enable_enhanced_contextEnable enhanced context gathering (uses more tokens but may improve accuracy)notrue
enable_ast_validationEnable AST-based validation to reduce false positives (recommended)notrue
use_structured_phasesEnable structured phased prompting to ensure files are edited before JSON generationnotrue
enable_educational_prEnable educational PR descriptions with security explanationsnotrue
executable_testsRFC-060: Enable executable test generation in VALIDATE phasenotrue
create_prCreate Pull Request after successful mitigation (default: true for production)notrue
pipeline_run_idRFC-126: Pipeline run ID from a prior scan step (required for mode=process)no""
namedescription
has_issuesWhether issues were found for processing
security_findingsSecurity vulnerabilities found in JSON format
educational_contentEducational explanation of the issues and fixes
issue_analysisFull analysis of the issues in JSON format
scan_resultsResults from vulnerability scan in JSON format
created_issuesGitHub issues created from scan results (JSON array)
issues_createdNumber of GitHub issues created from scan results
issue_numbersFlat JSON array of issue numbers for GitHub Actions matrix strategy (e.g., [230,231,232])
pipeline_run_idRFC-126: Pipeline run ID (set by scan and process modes for downstream steps)