Provadyne runs the latest C coding-standard checks on your own machine — with AI-suggested fixes, SARIF for GitHub code scanning, and auditor-ready reports. No cloud upload. No procurement cycle.
#include <stdlib.h>int g = 0;void arm(int *p) { *p = 5;p. if (g = 1) { /* ... */ }}No agents to install on a build server, no code leaving your network.
One binary, no daemon farm. brew install provadyne-daemon — or grab the release for macOS, Linux, or Windows.
In your editor, inline on GitHub PRs, or in CI. Pick any mix of MISRA C:2023, CERT C, and CWE per run.
Apply AI-suggested fixes, upload SARIF to code scanning, and export auditor-ready PDF / CSV reports.
Powered by widely-used open source (cppcheck, fpdf2). No telemetry, no SaaS lock-in.
MISRA C:2023, SEI CERT C, and CWE in a single local pass — every finding tagged with its standard and CWE id so you filter by what your program requires.
One-click diff previews for each violation. Bring your own key (OpenAI / Anthropic) or run fully offline with a local Ollama model.
Emit SARIF 2.1.0 and upload it to GitHub code scanning — violations show up inline on the PR and under the Security tab, CWEs included.
Turn a directory of .c files into a batch PDF with severity rollups and per-file rule citations. CSV / Excel export included. Pro
Link Git history to ALM tickets (Codebeamer, Jira) and build a V-model requirement → code → test matrix — the coverage evidence auditors ask for. Pro
The daemon runs on localhost and works with no network at all. Your source never leaves the machine unless you opt into a BYOK API key.
Drop the GitHub Action into a workflow. It installs the engine, runs the checks, fails on the severity you choose, and uploads SARIF.
fail-on gate: mandatory / required / advisory / none# Detect, then upload to code scanning jobs: misra: runs-on: ubuntu-latest permissions: security-events: write steps: - uses: actions/checkout@v4 - id: scan uses: sgchoi/provadyne-action@v1 with: paths: 'src/**/*.c' output-format: sarif - uses: github/codeql-action/upload-sarif@v3 with: sarif_file: ${{ steps.scan.outputs.sarif-path }}
No seat counting, no procurement cycle. Pay by card; get a license file by email in minutes.
localhost and analysis is fully local. Nothing is uploaded unless you explicitly enable a BYOK API key for AI suggestions — and even then only the relevant snippet is sent to your chosen provider.