🔍 Article Relevance Audit — Week of 2026-03-27

[github-actions]

Articles scanned: 49 | New this week: 11 | Total findings: 15

---

🔗 Link Health

Articles scanned: 49 | Links checked: ~80 | Issues found: 3

Broken Internal Links

Article	Link Text	Broken URL	HTTP Status	Suggested Fix
agentic-ai-transforming-dev-teams	"GitHub Copilot CLI worktrees"	/articles/copilot-fleet-mode-video-pipeline	404 — no article with this slug	Change to /articles/video-pipeline-with-fleet-mode ⚠️ 3rd consecutive audit unfixed

Action: src/content/articles/agentic-ai-transforming-dev-teams.mdx, line 31. Replace /articles/copilot-fleet-mode-video-pipeline with /articles/video-pipeline-with-fleet-mode.

Unreachable External Links (Persistent)

Article	URL	Status	Notes
git-worktree-unlocks-agentic-development	`(resources.anthropic.com/redacted)	Unreachable	3rd audit. Possible PDF gating or wrong URL. Consider removing or replacing.
agent-harnesses-controlling-ai-agents-2026	`(www.anthropic.com/redacted)	Unreachable	3rd audit. Search for correct URL or remove citation.

---

📊 Factual Freshness

Articles scanned: 49 | Stale references found: 5

Potentially Stale References

Article	What It Says	Current Reality	Severity	Suggested Update
ai-fixes-its-own-bugs	"SWE-agent solves 12.47% of real-world bugs on SWE-bench"	SWE-bench top models now solve 50%+ of issues; 12.47% was GPT-4 performance in 2023–2024 context, now severely outdated	🔴 High	Add context: "At time of publication, SWE-agent's GPT-4 baseline solved 12.47%. As of 2026, top-performing systems exceed 50% on the same benchmark — the field has moved fast."
ai-fixes-its-own-bugs	"CodeQL Autofix powered by GPT-4o"	GitHub regularly updates the underlying model; GPT-4o was current at publication (Feb 2026) but may have been updated	🟡 Medium	Rephrase to avoid naming the specific model, or verify current model from GitHub docs
copilot-developer-fulfillment	"20 million GitHub Copilot users"	GitHub has reported higher user counts; 20M was the figure circa late 2024	🟡 Medium	Check GitHub's current user count announcement and update. LinkedIn posts from GitHub suggest it may be 25M+ by now.
choosing-the-right-ai-sdk	"GitHub Copilot SDK: Technical Preview (Jan 2026)"	Published Feb 2026; 2.5 months have passed — SDK may have gone GA	🟡 Medium	Check https://github.blog/changelog/?s=copilot+sdk for GA announcement; update status table row
az-prototype-scaffold-enterprise-azure-with-copilot	"Stable release is still tracking for next week" (written Mar 23)	As of Mar 27, latest release is still v0.2.1b5 (beta) — the "next week" window has passed without a stable release	🟡 Medium	Update once stable releases; monitor https://github.com/Azure/az-prototype/releases

---

🔀 Cross-Linking Opportunities

Opportunities found: 8 new articles with zero inbound links from related older articles.

---

context-engineering-key-to-ai-development

File: src/content/articles/context-engineering-key-to-ai-development.mdx
Published: 2026-02-14 | Missing links to newer related articles:

1. Link to: karpathy-directs-ai-agents-december-shift (2026-03-23)

   • Suggested section: "## The Most Underrated Skill in Modern Development"
   • Insert near: "When [Andrej Karpathy]((pureai.com/redacted) — co-founder of OpenAI — publicly stated '+1 for context engineering'..."
   • Suggested text: By December 2025, Karpathy had gone even further — [he stopped writing code entirely and went 100% agentic](/articles/karpathy-directs-ai-agents-december-shift), describing the shift as "unrecognizable." That's not hyperbole. It's context engineering at full scale.

2. Link to: github-spec-kit-english-to-production-specs (2026-03-23)

   • Suggested section: "## The Three Pillars of Context Engineering" or "## Practical Playbook"
   • Insert near: the discussion of instruction files and grounding documents
   • Suggested text: GitHub's [Spec-Kit](/articles/github-spec-kit-english-to-production-specs) is one of the best implementations of this principle I've seen — it turns English requirements into structured spec artifacts that AI agents use as context throughout the development cycle.

---

agentic-ai-transforming-dev-teams

File: src/content/articles/agentic-ai-transforming-dev-teams.mdx
Published: 2026-02-20 | Missing links to newer related articles:

1. Link to: karpathy-directs-ai-agents-december-shift (2026-03-23)

   • Suggested section: "## The Biggest Mindset Shift: Developers Are Product Owners Now"
   • Insert near: "if you're a developer in the agentic AI world, you're not just a developer anymore. You're a product owner"
   • Suggested text: The starkest data point I've seen: [Andrej Karpathy went from writing some code to writing zero code — 100% agent-directed — since December 2025](/articles/karpathy-directs-ai-agents-december-shift). Not gradually. Suddenly. That's the trajectory.

2. Link to: copilot-coding-agent-50-percent-faster-session-visibility (2026-03-23)

   • Suggested section: "## The Circle Replaces the Hierarchy"
   • Insert near: "GitHub recently shipped their Copilot Coding Agent in record time..."
   • Suggested text: Change inline mention of Copilot Coding Agent to also reference: GitHub recently shipped their Copilot Coding Agent — and then [shipped a 50% startup speed improvement just weeks later](/articles/copilot-coding-agent-50-percent-faster-session-visibility). That's the pace of iteration this new model enables.

3. Link to: leadership-direction-adopting-agentic-platforms (2026-03-23)

   • Suggested section: "## What This Means for Teams Right Now"
   • Insert near: the list of actions for engineering leaders
   • Suggested text: I wrote a dedicated piece for leaders navigating this: [Your Agentic Platform Choice Is Invisible Risk — Until It's Too Late](/articles/leadership-direction-adopting-agentic-platforms) covers the Gartner data, governance considerations, and how to structure the decision.

---

sandboxes-missing-infrastructure-layer-agentic-devops

File: src/content/articles/sandboxes-missing-infrastructure-layer-agentic-devops.mdx
Published: 2026-03-19 | Missing links to newer related articles:

1. Link to: nvidia-openshell-sandbox-ai-agents (2026-03-23)
   • Suggested section: "## NVIDIA OpenShell: Policy-Driven Agent Sandboxing"
   • Insert near: The section already covers OpenShell but doesn't link to the dedicated deep-dive
   • Suggested text: Add at start of section: I covered OpenShell in depth in [a standalone article](/articles/nvidia-openshell-sandbox-ai-agents) — including the PR #476 contribution, YAML policy examples, and comparison with E2B and Daytona. Here's the quick version:

---

vibe-testing-when-ai-agents-goodhart-your-test-suite

File: src/content/articles/vibe-testing-when-ai-agents-goodhart-your-test-suite.mdx
Published: 2026-03-08 | Missing links to newer related articles:

1. Link to: research-plan-implement-anti-vibe-coding-workflow (2026-03-24)
   • Suggested section: "## The Agentic DevOps Response to Vibe Testing"
   • Insert near: closing paragraph — "Vibe coding gave us code nobody understands. Vibe testing gives us confidence nobody earned."
   • Suggested text: If you want to see this workflow built end-to-end with GitHub Copilot CLI and agentic workflows, [I built and documented the full implementation](/articles/research-plan-implement-anti-vibe-coding-workflow) — including the Research → Plan → Implement loop that enforces quality without babysitting agents.

---

self-healing-infrastructure-with-agentic-ai

File: src/content/articles/self-healing-infrastructure-with-agentic-ai.mdx
Published: 2026-02-18 | Missing links to newer related articles:

1. Link to: multi-service-deployments (2026-03-23)
   • Suggested section: "## What to Look Into Next"
   • Insert near: the tools and next-steps section
   • Suggested text: For multi-service coordination challenges — where the healing of one service needs to account for its dependents — I covered [multi-service deployment orchestration patterns](/articles/multi-service-deployments) in a follow-up piece.

---

copilot-cli-biggest-week-yet

File: src/content/articles/copilot-cli-biggest-week-yet.mdx
Published: 2026-02-14 | Missing links to newer related articles:

1. Link to: copilot-cli-remote-access-your-agent-from-anywhere (2026-03-23)
   • Suggested section: "## The Terminal Is Becoming First-Class"
   • Insert near: closing section about platform direction
   • Suggested text: That platform direction continued: the --remote flag — [access your Copilot CLI agent from anywhere, including your phone](/articles/copilot-cli-remote-access-your-agent-from-anywhere) — shipped a few weeks later and made the "terminal as platform" vision explicit.

---

🌿 Evergreen Maintenance

Refresh candidates found: 3 high-priority

Refresh Candidates

Article	Published	Last Updated	Topic Area	Why Refresh?	Priority
context-engineering-key-to-ai-development	2026-02-14	Never	Context Engineering	GitHub Spec-Kit, az-prototype's neural embedding governance, Karpathy's December shift — all directly extend or validate the pillar article's thesis with real 2026 developments	🔴 High
agentic-ai-transforming-dev-teams	2026-02-20	Never	Agentic AI / Teams	Has a broken internal link (3rd audit); Karpathy's 100% agentic shift, copilot coding agent improvements, and leadership platform risk data all materially update the article	🔴 High
ai-fixes-its-own-bugs	2026-02-14	Never	Autonomous Code Repair	SWE-bench leaderboard has moved from 12.47% to 50%+ at the top; a new section on current benchmarks is needed; two stale factual claims in this article	🟡 Medium

Detailed Refresh Suggestions

context-engineering-key-to-ai-development

File: src/content/articles/context-engineering-key-to-ai-development.mdx

• What changed since publication:
  • Karpathy publicly described going 100% agent-directed since December 2025 — the most prominent real-world example of the article's thesis
  • GitHub Spec-Kit (Mar 2026) operationalizes context engineering for spec-driven development
  • az-prototype's neural embedding governance system (Mar 2026) is a textbook example of the "Context Management" pillar — compressing 40KB of governance policies to 1–2KB with embeddings
• Suggested updates:
  • Add Karpathy's December shift as a concrete example in the opening section (internal link ready: karpathy-directs-ai-agents-december-shift)
  • Add GitHub Spec-Kit to "Practical Playbook" as a tool that embodies context engineering (internal link: github-spec-kit-english-to-production-specs)
  • Add az-prototype's policy embedding approach as a "Context Management" example (internal link: az-prototype-scaffold-enterprise-azure-with-copilot)
• Sources: /articles/karpathy-directs-ai-agents-december-shift, /articles/github-spec-kit-english-to-production-specs, /articles/az-prototype-scaffold-enterprise-azure-with-copilot

agentic-ai-transforming-dev-teams

File: src/content/articles/agentic-ai-transforming-dev-teams.mdx

• What changed since publication:
  • Broken internal link (/articles/copilot-fleet-mode-video-pipeline) — fix to /articles/video-pipeline-with-fleet-mode
  • Karpathy's December shift is the strongest evidence yet for the article's "developers as product owners" thesis
  • Copilot coding agent 50% speed improvement shows how fast the tooling is maturing
  • Leadership platform risk (Gartner: 40% of agentic AI projects cancelled by 2027) adds a governance dimension
• Suggested updates:
  • Fix the broken internal link
  • Add Karpathy paragraph in "## The Biggest Mindset Shift" section
  • Add Copilot coding agent speed improvement near existing Copilot Coding Agent mention
  • Link to leadership-direction-adopting-agentic-platforms in "## What This Means for Teams Right Now"

ai-fixes-its-own-bugs

File: src/content/articles/ai-fixes-its-own-bugs.mdx

• What changed since publication:
  • SWE-bench leaderboard has moved dramatically — top systems now solve 50%+ of issues vs. 12.47% (GPT-4 baseline, NeurIPS 2024)
  • CodeQL Autofix model may have been updated by GitHub
• Suggested updates:
  • Add a note after the 12.47% claim contextualizing current leaderboard state
  • Verify and update the CodeQL Autofix model reference
  • Consider adding a new "Where the Benchmark Stands Today" paragraph in the SWE-agent section
• Sources: [SWE-bench leaderboard]((www.swebench.com/redacted), 2025 leaderboard analysis

Generated by Article Relevance Audit — Weekly Health Check · ◷

[github-actions]

🤖 Auto-triage by issue-worker

Why this is suitable for Copilot:
This issue contains highly specific, mechanical tasks — an exact 1-line broken link fix with the precise file path and line number, plus 8 cross-linking opportunities with exact suggested markdown text, target section names, and anchor context. No editorial judgment required to execute these changes.

What Copilot should do:

Priority 1 — Broken internal link fix (1 line change):

• File: src/content/articles/agentic-ai-transforming-dev-teams.mdx, line 31
• Change: /articles/copilot-fleet-mode-video-pipeline → /articles/video-pipeline-with-fleet-mode
• ⚠️ This has been flagged in 3 consecutive audits and must be fixed

Priority 2 — Cross-linking opportunities (exact markdown provided):
For each of the following, insert the suggested text near the specified anchor text in the indicated section:

1. context-engineering-key-to-ai-development.mdx → link to karpathy-directs-ai-agents-december-shift (section: "## The Most Underrated Skill in Modern Development")
2. context-engineering-key-to-ai-development.mdx → link to github-spec-kit-english-to-production-specs (section: "## The Three Pillars of Context Engineering" or "## Practical Playbook")
3. agentic-ai-transforming-dev-teams.mdx → link to karpathy-directs-ai-agents-december-shift (section: "## The Biggest Mindset Shift")
4. agentic-ai-transforming-dev-teams.mdx → update inline Copilot Coding Agent mention to link to copilot-coding-agent-50-percent-faster-session-visibility (section: "## The Circle Replaces the Hierarchy")
5. agentic-ai-transforming-dev-teams.mdx → link to leadership-direction-adopting-agentic-platforms (section: "## What This Means for Teams Right Now")
6. sandboxes-missing-infrastructure-layer-agentic-devops.mdx → add link to nvidia-openshell-sandbox-ai-agents at start of NVIDIA OpenShell section
7. vibe-testing-when-ai-agents-goodhart-your-test-suite.mdx → link to research-plan-implement-anti-vibe-coding-workflow near closing paragraph in "## The Agentic DevOps Response to Vibe Testing"
8. self-healing-infrastructure-with-agentic-ai.mdx → link to multi-service-deployments in "## What to Look Into Next"
9. copilot-cli-biggest-week-yet.mdx → link to copilot-cli-remote-access-your-agent-from-anywhere in "## The Terminal Is Becoming First-Class"

The exact suggested markdown text for each insertion is in the issue body above.

Skip for now: Factual freshness items (SWE-bench update, user count check, SDK status) require external verification — leave those for human review.

Confidence: High

Generated by Issue Worker — Backlog Triage Agent · ◷

[github-actions]

🤖 Auto-triage by issue-worker

Why this is suitable for Copilot:
This audit contains a persistent broken internal link with an exact 1-line fix (3rd consecutive audit it appears unfixed), plus 8 well-scoped cross-linking opportunities — each with specific file paths, target sections, and suggested markdown text ready to insert. The mechanical nature of these changes makes them ideal for automated resolution.

What Copilot should do:

1. Broken link fix (highest priority — 3rd consecutive audit):

   • File: src/content/articles/agentic-ai-transforming-dev-teams.mdx, line 31
   • Replace: /articles/copilot-fleet-mode-video-pipeline → /articles/video-pipeline-with-fleet-mode

2. Cross-linking additions (use the exact suggested markdown from the issue body):

   • context-engineering-key-to-ai-development.mdx: add 2 links (Karpathy article in "## The Most Underrated Skill" section; GitHub Spec-Kit in "## The Three Pillars" section)
   • agentic-ai-transforming-dev-teams.mdx: add 3 links (Karpathy in "## The Biggest Mindset Shift"; Copilot coding agent speed near existing Copilot mention; leadership article in "## What This Means for Teams Right Now")
   • sandboxes-missing-infrastructure-layer-agentic-devops.mdx: add 1 link to OpenShell deep-dive at start of OpenShell section
   • vibe-testing-when-ai-agents-goodhart-your-test-suite.mdx: add 1 link to anti-vibe workflow article in closing paragraph
   • self-healing-infrastructure-with-agentic-ai.mdx: add 1 link to multi-service deployments in "## What to Look Into Next"
   • copilot-cli-biggest-week-yet.mdx: add 1 link to remote access article in closing section

Skip for now: Unreachable external links (require external research), factual freshness items (require external verification), and evergreen refresh suggestions (require editorial judgment).

Confidence: High

Generated by Issue Worker — Backlog Triage Agent · ◷

[github-actions]

🤖 Auto-triage by issue-worker

Why this is suitable for Copilot:
This audit contains a confirmed-unfixed broken internal link (verified still present in the codebase today — 4th consecutive audit flagging it) plus 8 cross-linking opportunities with exact file paths, target section headings, and ready-to-insert markdown text. No editorial judgment required.

What Copilot should do:

Priority 1 — Broken internal link (confirmed present, fix this first):

• File: src/content/articles/agentic-ai-transforming-dev-teams.mdx, line 31
• Replace: /articles/copilot-fleet-mode-video-pipeline → /articles/video-pipeline-with-fleet-mode
• ⚠️ This has been unfixed across 4 consecutive audits

Priority 2 — Cross-linking (use exact suggested text from issue body):

1. context-engineering-key-to-ai-development.mdx → link to karpathy-directs-ai-agents-december-shift in "## The Most Underrated Skill in Modern Development" section
2. context-engineering-key-to-ai-development.mdx → link to github-spec-kit-english-to-production-specs in "## The Three Pillars of Context Engineering" section
3. agentic-ai-transforming-dev-teams.mdx → link to karpathy-directs-ai-agents-december-shift in "## The Biggest Mindset Shift" section
4. agentic-ai-transforming-dev-teams.mdx → update Copilot Coding Agent mention to link to copilot-coding-agent-50-percent-faster-session-visibility in "## The Circle Replaces the Hierarchy"
5. agentic-ai-transforming-dev-teams.mdx → link to leadership-direction-adopting-agentic-platforms in "## What This Means for Teams Right Now"
6. sandboxes-missing-infrastructure-layer-agentic-devops.mdx → add link to nvidia-openshell-sandbox-ai-agents at start of OpenShell section
7. vibe-testing-when-ai-agents-goodhart-your-test-suite.mdx → link to research-plan-implement-anti-vibe-coding-workflow in closing paragraph of "## The Agentic DevOps Response to Vibe Testing"
8. self-healing-infrastructure-with-agentic-ai.mdx → link to multi-service-deployments in "## What to Look Into Next"
9. copilot-cli-biggest-week-yet.mdx → link to copilot-cli-remote-access-your-agent-from-anywhere in "## The Terminal Is Becoming First-Class"

Skip: Unreachable external links, factual freshness items, and evergreen refresh suggestions — those require external verification.

Confidence: High

Assigning to Copilot coding agent.

Generated by Issue Worker — Backlog Triage Agent · ◷