Search
Results for “ai-agents”
- software-engineering
Where to Put Human Approval Gates in AI-Powered Developer Workflows
Let an agent prepare release candidates, run checks, and explain its choices; pause before publishing or changing shared state. A worked GitHub Actions and npm example shows how to make the boundary reviewable and enforceable.
- engineering
How to Monitor a Feature Rollout Before Trusting Its AI Kill Switch
A rollout gate and a runtime kill switch solve different problems. Use this checklist to set measurable exposure limits, compare candidate and control behavior, and decide exactly what an AI kill switch should change.
- productivity
Build a Faster Technical Knowledge Feed With AI Curation, Tags, and Weekly Reviews
Turn an AI-driven news feed into a practical technical reading system. This workflow shows how to filter articles, save only reusable ideas, tag them consistently, and convert a week of reading into decisions and experiments.
- software-engineering
AI Code Validation vs Release Governance: 9 Production Pipeline Checks to Add First
AI can assess changes and recommend whether a release is safe, but it should not silently become the authority that approves production. Use this control checklist to assign decision ownership, preserve audit evidence, route security review, and handle model-policy conflicts.
- software-engineering
GitHub Copilot Agent Mode vs Copilot Edits in VS Code: A Task-Selection Workflow for Safer Changes
Use Copilot Edits for bounded changes you can specify and review file by file. Use agent-oriented workflows when a task requires discovery, implementation, testing, and iteration—and add explicit checkpoints before generated changes reach a branch.
- ci-cd
A Reviewer Playbook for AI-Generated GitHub Actions Workflows
AI-generated GitHub Actions YAML can be syntactically correct while creating a direct path from a forked pull request to repository credentials. Use this workflow-specific review playbook to inspect permissions, action pins, triggers, interpolation, secrets, and checkout context before merging.
- software-engineering
A 10-Point Review Rubric for AI-Generated Pull Requests That Catch Architectural Debt
AI-generated pull requests can pass every visible test while quietly creating duplicate policy, unsafe boundaries, and unsupported dependencies. Use this five-gate, 10-point rubric to decide what can merge, what needs follow-up, and what needs an architectural redesign.
- software-engineering
A Pre-Launch GenAI Risk Checklist for Internal Chatbots and Developer Tools
Turn NIST’s generative AI risk guidance into a practical release gate for a coding assistant, repository chatbot, or internal support bot. Use this checklist to test data exposure, harmful outputs, human review, evaluation coverage, and incident response before enabling real users.
