Updates

Operating Principles

• Evidence first: Record source publication dates and read dates.
• Interpretation first: Track how new information changes the handbook's view.
• Separate structure and interpretation: Distinguish editing changes from conceptual changes.
• Maintain a watchlist: Do not rely on one person's update memory.

External Sources to Watch

Change Log

2026-06-13

Structure Changes

• Review and refresh round: fixed a locale-leaking cross-read link in index.mdx (bare /books/agent-orchestration-patterns → /ko/books/agent-orchestration-patterns, annotated as Korean) so it no longer 308-redirects out of the English edition.
• Bumped index.mdx updatedAt to 2026-06-13.

Interpretation Changes

• None. No conceptual or interpretive change to the handbook in this round.

Affected Chapters

• index.mdx

Evidence Links

• N/A (editorial/link-hygiene change only)

2026-05-23

Structure Changes

• Created English edition under apps/handbook/content/books/en/harness-engineering.
• Added 23 English MDX pages and meta.json.
• Added English home catalog override for harness-engineering.
• Reflected OpenAI Agents SDK, Codex remote/hooks, Secure MCP Tunnel, and OpenAI Developers plugin.
• Reflected Anthropic Managed Agents, Claude Code auto mode, and financial agents.
• Reflected latest gstack and revfactory/harness README interpretation from the Korean baseline update.
• Added verification and update appendix pages.

Interpretation Changes

• Previous view: a harness is a team work system made of repo structure, docs, evaluation, approval, and operations.
• Updated view: that remains true, but 2026 product changes show more harness elements becoming standard primitives: MCP, skills, sandbox, hooks, remote approvals, open-source harnesses, plugins, session logs, and permission classifiers.
• Additional view: Anthropic Managed Agents and auto mode move harness design beyond evaluation loops into durable session, stateless harness, sandbox/hands, credential boundary, and classifier policy design.
• Additional view: Anthropic financial agents and OpenAI Developers plugin show domain templates and provider setup becoming rollout surfaces.

Affected Chapters

• index.mdx
• foundations.mdx
• repo-readable-systems.mdx
• five-elements.mdx
• engineering-mechanics.mdx
• evaluation-loops.mdx
• case-studies.mdx
• case-openai.mdx
• case-anthropic.mdx
• case-toss.mdx
• case-gstack.mdx
• case-revfactory.mdx
• domain-playbooks.mdx
• scenario-frontend-team.mdx
• scenario-platform-team.mdx
• scenario-payments-team.mdx
• scenario-ai-product-team.mdx
• make-it-yours.mdx
• team-rollout.mdx
• checklist.mdx
• operations.mdx
• verification.mdx
• updates.mdx
• apps/handbook/lib/home-server.ts

Evidence Links

• https://openai.com/ko-KR/index/harness-engineering/
• https://openai.com/index/the-next-evolution-of-the-agents-sdk/
• https://openai.com/index/work-with-codex-from-anywhere/
• https://developers.openai.com/api/docs/changelog
• https://developers.openai.com/api/docs/guides/agents
• https://developers.openai.com/api/docs/guides/agents/sandboxes
• https://developers.openai.com/codex/hooks
• https://developers.openai.com/codex/remote-connections
• https://developers.openai.com/learn/developers-codex-plugin
• https://www.anthropic.com/engineering/harness-design-long-running-apps
• https://www.anthropic.com/engineering/claude-code-auto-mode
• https://code.claude.com/docs/en/permission-modes
• https://code.claude.com/docs/en/auto-mode-config
• https://www.anthropic.com/engineering/managed-agents
• https://platform.claude.com/docs/en/managed-agents/overview
• https://platform.claude.com/docs/en/managed-agents/mcp-connector
• https://www.anthropic.com/news/finance-agents
• https://toss.tech/article/harness-for-team-productivity
• https://github.com/garrytan/gstack
• https://github.com/revfactory/harness

Template

YYYY-MM-DD

Structure Changes

• ...

Interpretation Changes

• Previous:
• Updated:
• Reason:

Affected Chapters

• ...

Evidence Links

• ...