Skip to content

Article Relevance Audit — Week of 2026-03-20 #56

@github-actions

Description

@github-actions

Weekly automated audit of all 38 evergreen articles. 2 new articles discovered since last run (2026-03-13): git-worktree-unlocks-agentic-development and sandboxes-missing-infrastructure-layer-agentic-devops (both 2026-03-19). github-copilot-cli-extensions-complete-guide (2026-03-14) was also new this cycle.

Articles scanned: 38 | Links checked: 553 external, 65 internal | Issues found: 9


🔗 Audit 1: Link Health

Internal links checked: 65 | Broken: 1 (persistent) | Unreachable: 2

Broken Internal Links

Article Link Text Broken URL Suggested Fix
agentic-ai-transforming-dev-teams "GitHub Copilot CLI worktrees" /articles/copilot-fleet-mode-video-pipeline Update to /articles/git-worktree-unlocks-agentic-development — the new article is a perfect match for this link text and intent. Previously suggested fix was /articles/video-pipeline-with-fleet-mode; the new article is a better target.

Context: Line 31 of agentic-ai-transforming-dev-teams.mdx:

"experiments get spun up using GitHub Copilot CLI worktrees or coding agents"

Replace the broken slug with git-worktree-unlocks-agentic-development — article title is "Git Worktree: The Infrastructure That Unlocks Agentic Development." This is a direct match. ✅

Unreachable External Links (firewall or server issue — verify manually)

Article URL Status
git-worktree-unlocks-agentic-development `(resources.anthropic.com/redacted) fetch failed (PDF may be gated or URL wrong)
agent-harnesses-controlling-ai-agents-2026 `(www.anthropic.com/redacted) unreachable since 2026-03-11 audit — persistent

📊 Audit 2: Factual Freshness

Articles scanned: 38 | Stale references found: 5 (all carry-overs from 2026-03-13 audit — not yet actioned)

Potentially Stale References

Article What It Says Current Reality Severity Suggested Update
ai-fixes-its-own-bugs "SWE-agent solves 12.47% of SWE-bench" Top models now solve 50%+ on SWE-bench Verified (as of NeurIPS 2025) 🔴 High Update stat with current leaderboard link; add context that the bar has moved dramatically
ai-fixes-its-own-bugs "CodeQL Autofix powered by GPT-4o" GitHub has iterated on underlying models since Feb 2024 blog post 🟡 Medium Soften to "AI-powered" or verify current model via GitHub docs
copilot-developer-fulfillment "20 million GitHub Copilot users" 30M+ users reported by late 2025/early 2026 🟡 Medium Update to 30M+ with a current source
copilot-developer-fulfillment "2024 DORA report" 2025 DORA report is now available 🟡 Medium Link to 2025 edition; check if findings align or diverge
choosing-the-right-ai-sdk "GitHub Copilot SDK: Technical Preview (Jan 2026)" ~10 weeks since launch — check if SDK has moved to GA 🟡 Medium Verify current status at https://github.com/github/copilot-sdk and update if promoted

New Articles — No Freshness Issues Found

git-worktree-unlocks-agentic-development, sandboxes-missing-infrastructure-layer-agentic-devops, and github-copilot-cli-extensions-complete-guide are all newly published and contain no obviously stale references.


🔀 Audit 3: Cross-Linking Opportunities

New opportunities found: 5 (3 for new articles + 2 persistent carryovers)

git-worktree-unlocks-agentic-development

File: src/content/articles/git-worktree-unlocks-agentic-development.mdx
Published: 2026-03-19 | Inbound links: 0 from older articles

This article is also the fix target for the broken link in audit 1. But older articles should proactively reference it:

  1. Link from: agentic-devops-next-evolution-of-shift-left

    • Suggested section: ## What Agentic DevOps Actually Looks Like
    • Insert near: "DevOps processes designed to enable your agents — not your humans, your agents"
    • Suggested text: The infrastructure piece that makes parallel agentic work actually feasible is [git worktree](/articles/git-worktree-unlocks-agentic-development) — it went from niche power feature to essential agentic tooling almost overnight.
  2. Link from: top-5-mistakes-creating-custom-github-copilot-agents

    • Suggested section: ## Mistake #3: Ignoring Context Window Reality
    • Insert near: "Context windows vary widely by model"
    • Suggested text: If you're running multiple agents in parallel — which you should be — [git worktree is the infrastructure that makes it possible](/articles/git-worktree-unlocks-agentic-development) without them stepping on each other's context.

github-copilot-cli-extensions-complete-guide

File: src/content/articles/github-copilot-cli-extensions-complete-guide.mdx
Published: 2026-03-14 | Inbound links: 1 (only from git-worktree-unlocks-agentic-development)

  1. Link from: top-5-mistakes-creating-custom-github-copilot-agents

    • Suggested section: ## Build With Discipline, Not Just Detail
    • Insert near: "The tooling is maturing fast. GitHub Copilot Extensions give you a platform."
    • Suggested text: If you're building on the CLI specifically, the [Copilot CLI extension system](/articles/github-copilot-cli-extensions-complete-guide) is the most powerful control surface nobody's talking about — custom tools, lifecycle hooks, and hot-reload governance in a single .mjs file.
  2. Link from: github-copilot-vs-the-world

    • Suggested section: wherever GitHub Copilot's extensibility is discussed
    • Suggested text: The CLI's [extension system](/articles/github-copilot-cli-extensions-complete-guide) is the deepest answer to "but can you customize it?" — full Node.js processes that participate in the agent loop, not just config scripts.

Persistent: Zero-Inbound Articles (still not linked from any evergreen article)

These were flagged in the 2026-03-13 audit and remain unaddressed:

Article Published Top Cross-Link Candidate
github-copilot-vs-the-world 2026-03-12 Link from choosing-the-right-ai-sdk closing section
stop-pulling-your-agentic-unicorns-off-the-field 2026-03-12 Link from copilot-developer-fulfillment
vibe-testing-when-ai-agents-goodhart-your-test-suite 2026-03-08 Link from tests-are-everything-agentic-ai
specs-equal-tests-terraform-ai-development 2026-03-06 Link from agent-proof-architecture-agentic-devops
self-healing-infrastructure-with-agentic-ai 2026-02-18 Link from agentic-devops-next-evolution-of-shift-left
cryptographic-approval-gates-ai-agents 2026-02-20 Link from agent-harnesses-controlling-ai-agents-2026
ai-audit-codebase-testability 2026-02-20 Link from context-engineering-key-to-ai-development
github-copilot-visual-studio-guide 2026-02-20 Link from building-the-future-with-ai
lessons-from-500-github-migrations 2026-02-14 Link from github-agentic-workflows-hands-on-guide
vidpipe-copilot-cli-challenge 2026-02-14 Link from introducing-vidpipe-ai-video-pipeline

🌿 Audit 4: Evergreen Maintenance

Refresh candidates: 5 (all carry-overs from 2026-03-13 audit — not yet actioned)

Article Published Last Updated Why Refresh? Priority
context-engineering-key-to-ai-development 2026-02-14 Never MCP now ubiquitous; new Copilot memory features; article predates git-worktree as a context strategy 🔴 High
ai-fixes-its-own-bugs 2026-02-14 Never SWE-bench leaderboard has radically changed; new self-healing patterns in agentic DevOps 🔴 High
choosing-the-right-ai-sdk 2026-02-14 Never Copilot SDK status may have changed; new SDKs in market since Jan 2026 🟡 Medium
building-the-future-with-ai 2026-02-13 Never Earliest article; 35+ articles now reference it as a pillar; deserves an update pass 🟡 Medium
agentic-ai-transforming-dev-teams 2026-02-20 Never Also has the broken internal link; git worktree pattern now mainstream; needs full refresh 🟡 Medium

Detailed Refresh Suggestions

context-engineering-key-to-ai-development

  • What changed: MCP (Model Context Protocol) is now the dominant standard for context injection; Copilot Memory shipped in VS Code 1.109; git worktree is now the recommended infrastructure for multi-agent context isolation
  • Add section on: MCP as the evolution of manual context engineering
  • Add cross-reference: my git worktree guide for multi-agent context isolation

ai-fixes-its-own-bugs

  • What changed: SWE-bench Verified leaderboard is dramatically different — top agents now solve 50%+, not 12%. The framing of the article needs a major data update.
  • Update: Replace the SWE-agent 12.47% stat with current leaderboard data; add section on agentic DevOps self-healing (sandboxes + harnesses make this safer)
  • Add cross-reference: sandboxes article for the infrastructure that makes autonomous bug-fixing safe

Automated audit run 2026-03-20 | Next run ~2026-03-27

Generated by Article Relevance Audit — Weekly Health Check ·

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions