diff --git a/config/previously-recommended.md b/config/previously-recommended.md
index 728a46b..e598dec 100644
--- a/config/previously-recommended.md
+++ b/config/previously-recommended.md
@@ -111,3 +111,12 @@ https://www.smashingmagazine.com/2025/07/design-patterns-ai-interfaces/
 https://mbrenndoerfer.com/writing/rag-architecture-retriever-generator-design-patterns
 https://www.youtube.com/watch?v=5ABcYSWPVr8
 >>>>>>> Stashed changes
+
+# Week 14 — 2026-07-12 (advanced from Week 13 — briefed 4x, pool exhausted; see run log)
+https://www.anthropic.com/engineering/writing-tools-for-agents
+https://www.smashingmagazine.com/2026/05/practical-interface-patterns-ai-transparency/
+https://www.oneusefulthing.org/p/management-as-ai-superpower
+https://www.anthropic.com/engineering/advanced-tool-use
+https://lukew.com/ff/entry.asp?2142=
+https://medium.com/@netmohan97/what-building-ai-agent-skills-taught-me-about-product-management-in-the-ai-era-a7235a250431
+https://simonwillison.net/2026/jul/3/judgement/
diff --git a/logs/run-2026-07-12.md b/logs/run-2026-07-12.md
new file mode 100644
index 0000000..24a6bc2
--- /dev/null
+++ b/logs/run-2026-07-12.md
@@ -0,0 +1,30 @@
+# Run Log — 2026-07-12
+Week: 14 (advanced from Week 13 — see note below)
+Topic: Week 14 · Agent-Computer Interface + sub-agents (tool naming grammar, decomposition over consolidation, progressive disclosure at the API layer)
+Sources checked: YouTube (13 attempted, all 403), Blogs/RSS (Tier 1 attempted, all 403; Anthropic mirror stale Nov 2025), Web (17 searches), X (skipped)
+Items collected: ~22 candidates surfaced via web search
+Items after hard filters: ~16 (excluded prior-recommended, checklist, and all 4 Week 13 briefings' URLs)
+Items after topic gate: ~12
+Must Read: 2 (score range: 88–93)
+Recommended: 3 (score range: 78–84) + 1 Discovered (68)
+Wild Card: yes (trigger: contrarian — sub-agent model choice, Simon Willison "Fable's judgement")
+Discovered: 1 (Netra Chandramohan, PM building agent skills)
+Dropped: several low-trust SEO agent-UX listicles that failed the quality bar
+Errors/Notes:
+- WEEK DETERMINATION OVERRIDE: The current-week algorithm resolves to Week 13 (first week with zero [x] items) ONLY because Katherine has not ticked any Week 13 boxes. Week 13 has already been briefed FOUR times (2026-06-14, 06-21, 06-28, 07-05) with ~24 distinct picks spanning the full multi-agent-orchestration space. Every prior briefing, with escalating urgency, recommended advancing to Week 14. A fifth Week 13 pass would violate the prompt's core "quality over quantity / content she hasn't read yet" principle. Idempotency guard (STEP 1d) did NOT trigger a stop — most recent Week 13 page was 2026-07-05 (7 days ago, outside the 3-day window). Decision: advanced focus to Week 14 with a transparent explanation at the top of the Notion page and a nudge to Katherine to tick Week 13 boxes so the checklist tracks reality.
+- Prompt file (prompts/weekly-briefing.md) and config files (sources.md, previously-recommended.md) contain unresolved git merge-conflict markers (Updated upstream / Stashed changes). Treated both sides as complementary/valid. These markers are committed content, not an active merge — worth resolving in a future cleanup.
+- Dedup log (previously-recommended.md) had not been updated since Week 7; the four Week 13 briefings' URLs were never logged. Dedup for this run was performed against the checklist + dedup log + all 4 Week 13 Notion pages directly. Appended Week 14 URLs to the dedup log this run.
+- Diversity rule satisfied: Product Lane ≥2 (Mollick, Chandramohan), Design Lane ≥2 (Smashing Part 2, Wroblewski), pure-engineer-audience ≤2 (Anthropic advanced tool use, Willison). Anthropic tool-design Highlight tagged "mixed."
+- Source Health: YouTube Failed (403), Blogs/RSS Failed (403 direct + proxy), Web OK, X Skipped. Datacenter-IP block, consistent with all prior cloud runs. Run carried entirely by web search.
+Duration: ~1 run (approx.)
+
+## Picks (7)
+1. ⭐ Anthropic — Writing effective tools for agents · 93 · https://www.anthropic.com/engineering/writing-tools-for-agents
+2. ⭐ Smashing — Practical Interface Patterns for AI Transparency (Part 2) · 88 · https://www.smashingmagazine.com/2026/05/practical-interface-patterns-ai-transparency/
+3. 🟠 Ethan Mollick — Management as AI superpower · 84 · https://www.oneusefulthing.org/p/management-as-ai-superpower
+4. 🟠 Anthropic — Advanced tool use on the Claude Developer Platform · 80 · https://www.anthropic.com/engineering/advanced-tool-use
+5. 🟠 Luke Wroblewski — Showing the Work of Agents in UI · 78 · https://lukew.com/ff/entry.asp?2142=
+6. 🔵 Netra Chandramohan — What Building AI Agent Skills Taught Me About PM · 68 · https://medium.com/@netmohan97/what-building-ai-agent-skills-taught-me-about-product-management-in-the-ai-era-a7235a250431
+7. 🃏 Simon Willison — Fable's judgement · 74 · https://simonwillison.net/2026/jul/3/judgement/
+
+Notion page: https://app.notion.com/p/39b82e5a5fa381988dc7f81c95491341