Agent payment safety infrastructure
Monarch Shield
A local-first CLI and CI safety gate for agent payment stacks. It detects supported money-moving code paths and fails strict checks when expected payment guards are missing.
The customer problem
Agent-payment code can be merged without an explicit preflight policy check, leaving teams to find an avoidable implementation gap late in review.
Turns agent-payment release risk into a deterministic preflight developers and coding agents can run before real funds move.
What I built
A product workflow that makes the next decision clearer.
- Focused on one detectable release-time failure mode rather than broad security claims.
- Built a command developers can run locally or in CI with strict failure behavior.
- Made findings deterministic and inspectable so teams can reproduce the result.
- Kept runtime payment authorization and settlement outside the tool's scope.
Working flow
From input to controlled action
- Scan a codebase for supported payment-code patterns
- Check for a detectable policy guard
- Report the finding with a reproducible command
- Fail CI in strict mode when the guard is missing
- Return control to the developer before release
Inspectable evidence
What works in the current build
- MIT-licensed public repository
- Installable CLI and CI-oriented strict mode
- Public evidence tied to shipped code
- Clear separation between build-time preflight and runtime enforcement
Operating boundary
What this case study does and does not claim
Monarch Shield is a build-time and CI preflight. It is not runtime enforcement by itself, fraud prevention, provider verification, wallet security, regulatory compliance, or settlement validation.
Available for open-source use. Related agent-payment workflow and integration work is available as a consulting engagement.
Relevant problem?
Build the version your customers can actually test.
Share the current workflow, user, and business constraint. I will tell you where I would start.
Discuss a related build