Chowkidaar
AI Reliability & Agentic Engineering
An API reliability layer that maps external-service drift to affected code paths, proposes focused fixes, runs checks, and opens PR-ready review flows.
By Shehryar Hassan·2026-09-18
FastAPIReactPythonTypeScriptOpenAIGitHub PRsGraph Analysis
github.com/shehryaur/ChowkidaarAPI
The Product VisionReplacing surprise integration breakages with early detection, focused remediation, and human-reviewed pull requests.
The User Problem
API integrations usually fail after a provider changes behavior and a developer notices in production. Teams lose time tracing blast radius before they can even write the fix.
Architecture
- Mapped external API usage into a repository graph so the agent works on affected call paths instead of wandering the whole codebase.
- Added simulations and live audits that score pressure, latency, failures, and recommended fixes before users feel an incident.
- Designed the apply workflow to write the smallest safe patch, run project checks, and open a PR review flow without auto-merging.
Product & Growth Impact
- Moves API maintenance from reactive debugging to proactive contract, environment, and traffic monitoring.
- Keeps secrets safer by fingerprinting credentials locally instead of storing raw values.
- Turns maintenance work into reviewable evidence: graph context, simulation output, validation checks, and a proposed branch.