🧰 GitHub Agentic Workflows

The items that name it 7

• AI Agent Hacking Writeups
  repoas affected products2026-09-17
• Top AI Coding Agent security resources — August 2026
  articleas affected products2026-08-03
• AI Security Briefing
  articleas affected products2026-07-18

  Feature in public preview, launched by GitHub in February; lets users write plain-English Markdown instructions to an AI agent that reads issues/PRs, runs tools, and replies; vulnerable to the GitLost technique when granted read access across an organization's repositories, including private ones

• AI Security Briefing
  writeupas affected products2026-07-18

  AI agent (backed by Claude or GitHub Copilot) paired with GitHub Actions; vulnerable to indirect prompt injection allowing cross-repository data leakage, including private repos

• JFrog: Managing AI Agent Primitives with APM
  articleas tools2026-07-15

  Runs AI agents as GitHub Actions in CI/CD, importing approved/pinned APM packages into a sandboxed runner with human approval gates and AI spend caps

• Agentic Workflows
  articleas tools2026-05-13

  research demonstrator from GitHub Next (announced February 2026 by Don Syme and Peli de Halleux) that lets users describe automation intent in Markdown with YAML frontmatter, compiled via gh aw compile into an inspectable GitHub Actions .lock.yml workflow

• Security Architecture of GitHub Agentic Workflows
  articleas tools2026-05-13

  Compiles authored workflows into GitHub Actions with explicit constraints (permissions, outputs, auditability, network access) to run agents safely