Carry context into the next session.
The application map and approved baseline remain available across conversations and coding assistants. Keep the approved rules with your project and review intentional permission changes alongside the code.
GALEN POLICY: HOW IT WORKS
Find the flaw.
Fix it before it ships.
Galen Policy maps your web application’s code paths, permissions, and data access. You or a coding agent with terminal access runs its standalone CLI. Review the access rules recovered from your code and approve the intended baseline. Later checks use that saved reference to find vulnerabilities and policy drift, with evidence to guide corrections.
Standalone command-line toolDeterministic analysisHuman-approved rules
SET THE BASELINE. KEEP BUILDING.
Set up once.
Check as the code changes.
Start with Galen and your coding agent in the same project environment. Then follow an administrator-only customer export from its approved rule to a refactor that removes a guard—and the evidence needed to put it back. Select a step to see what happens.
Initial setup · Step 1 of 7
Put Galen where your agent can run it.
Download the Galen Policy build for your operating system and make the CLI runnable in your development environment. Open your web application’s repository in your coding agent with file and terminal access. Confirm that the agent can run Galen against that project.
What you getThe project and Galen CLI are accessible in the same working environment.
EXAMPLE REQUEST TO YOUR CODING AGENT
Start in your application’s repository.
Use the Galen Policy CLI installed in this environment. Check the release instructions for the correct invocation. Run its initial analysis against this repository. Show me the recovered access map and any findings. Wait for my review before approving the policy baseline.
This is an example prompt, not a shell command. Galen performs the analysis; your coding agent runs the tool and helps explain its output.
A remote or cloud agent needs Galen installed in its own execution environment. A download on your laptop alone does not make it available there.
Illustrative walkthrough · Not live analysis or actual product output.
YOUR WORKFLOW / YOUR CONTROL
The agent makes the edit.
Galen checks the result.
Galen derives findings from code facts and verifies their derivation. Your coding agent receives the relevant rule, affected path, and missing control as a focused task, then uses that evidence to propose a correction.
You review the change and run Policy again. A deliberate change to the rules gets explicit approval, so the baseline stays aligned with your decisions.
MAKE IT PART OF THE WAY YOU BUILD
One baseline.
Many code changes.
Decide when the check runs.
Run the CLI yourself, or instruct your coding agent to run it after meaningful changes and before release. Record the verified Galen invocation and baseline location in your coding agent’s project instructions so later sessions have the same guidance. An instruction file guides the agent; it does not create an automatic check on every edit.
Review the new evidence.
A clean result means no findings within the supported analysis. Coverage and evidence gaps still matter; keep your tests and code review in the loop.
START WITH YOUR NEXT CHANGE
Build with your agent.
Check with Galen Policy.
Join the free beta, download Galen Policy for your system, and run your first analysis in your application’s repository.