Bugb
Documentation for guardlink, cxg and Bravos: three security tools you run yourself, on your own machines.
Every command on these pages was executed against a real build and its actual output pasted in. What you read is what the binary does, not what its source suggests it should.
guardlink and cxg are open source and free to run. Bravos runs its research loop free on your own machine across any number of repositories, each on its own, with no account. The code graph, SAST, multi-repository workspaces, CI, scheduling, pre-PR gates and campaigns are paid, with a 30-day no-card trial from the account portal. The npm CLI works through your coding agent and git.
- guardlink
Keep a threat model in the repository it describes. Annotations anchor to real code positions, so the model changes when the code does. Validates and diffs in CI, and serves the model to a coding agent over MCP.
Install guardlinkAnnotate a repository
2.7.1MITon npm - cxg
Scan a running target for vulnerabilities. Its checks are ordinary programs rather than pattern files, so a check can chain requests, compute, and parse a format before it decides. Every finding keeps the request and the response behind it.
Install cxgRun your first scanCLI reference
1.9.2Apache-2.0released - Bravos
Make the coding agent you already use work your threat model, code graph, SAST and authorized pentests, with the evidence kept in your repository. Start it from your PRDs, design documents and earlier threat models, kept in the repository for every agent to work from. The research loop is free across any number of independent repositories; linked multi-repository workspaces, gates, CI, SAST and the code graph are paid, with a 30-day no-card trial.
Install BravosBuild a threat modelContext from your design documentsFree and Team plans
1.18.0on npm

