TECHNICAL INTELLIGENCE BRIEF

1Executive Snapshot

• Agent harness: 30 dev-web/HN + 64 repo signals → CTO nên đo bằng task nội bộ, không tin leaderboard đơn lẻ.
• CLI/IDE agents: 6 product baselines (Claude Code/Codex/Cursor/Copilot/Devin/OpenCode proxy) → NEXA adapter cần vendor-neutral.
• Context engineering: Serena/ctx/Repowise/Mneme signals → FARE ưu tiên repo memory + architectural rules.
• Terminal eval: Dirac 393 pts/148 comments; Terminal-Bench-3 200 stars/234 forks → thêm 20 task terminal Fabbi.
• Governance: X unauth direct N/A, FB 0 usable → confidence medium, nhưng social completeness 3/4 đạt.

4CTO Evaluation Matrix

5Repo Watch

6Impact Coverage

7CTO Recommendations

8Must-read Sources / Source Appendix

1. S01 [dev_web] I built an agentic coding harness across three CLI hosts — 2 pts / 0 comments; homescout; HN dev discourse proxy
2. S02 [dev_web] Peers – Multi-agent AI coding with measurable convergence — 1 pts / 0 comments; dash0r; HN dev discourse proxy
3. S03 [dev_web] Show HN: Mneme HQ – repo-native architectural rules for AI coding agents — 1 pts / 0 comments; Tval; HN dev discourse proxy
4. S04 [dev_web] Aming Claw – Zero-orchestration multi-agent coding — 1 pts / 0 comments; aming557; HN dev discourse proxy
5. S05 [dev_web] Show HN: Unspaghettit – executable behavior specs for AI coding agents — 5 pts / 0 comments; D3F; HN dev discourse proxy
6. S06 [dev_web] Bill Gates AI on AI (one month later) — 3 pts / 0 comments; vbutsomesayw; HN dev discourse proxy
7. S07 [dev_web] Show HN: Simple Sprite Sheet Generation — 3 pts / 0 comments; armcat; HN dev discourse proxy
8. S08 [dev_web] Show HN: My first app, artisanally vibe-coded in 4 months — 3 pts / 4 comments; jeroen_stulen; HN dev discourse proxy
9. S09 [dev_web] Zero – Programming Language for Agents — 3 pts / 0 comments; xendo; HN dev discourse proxy
10. S10 [dev_web] Show HN: opub, donated compute for open-source — 2 pts / 0 comments; goodroot; HN dev discourse proxy
11. S11 [dev_web] Show HN: GoPOSIX – a Go-native POSIX userland, ~97% BusyBox-compatible — 2 pts / 0 comments; ramayac; HN dev discourse proxy
12. S12 [dev_web] Implicit Knowledge Is a Liability — 1 pts / 0 comments; gruyaume; HN dev discourse proxy
13. S13 [dev_web] Ask HN: Is agent-driven QA a thing? — 1 pts / 1 comments; straydusk; HN dev discourse proxy
14. S14 [dev_web] Ask HN: May be a basic question, but how can I use AI well? — 10 pts / 5 comments; jdw64; HN dev discourse proxy
15. S15 [dev_web] Launch HN: Kampala (YC W26) – Reverse-Engineer Apps into APIs — 100 pts / 83 comments; alexblackwell_; HN dev discourse proxy
16. S16 [dev_web] Ask HN: Opus 4.7 – is anyone measuring the real token cost on agentic tasks? — 1 pts / 0 comments; nicola_alessi; HN dev discourse proxy
17. S17 [dev_web] Show HN: Repowise – Codebase intelligence for AI coding agents (open source) — 1 pts / 0 comments; raghavchamadiya; HN dev discourse proxy
18. S18 [dev_web] Show HN: Salacia – The First Runtime OS for Agentic Coding — 1 pts / 1 comments; alfredhua; HN dev discourse proxy
19. S19 [dev_web] Show HN: Tracecore: Benchmark AI Agents on Deterministic Coding Tasks — 1 pts / 0 comments; extra_cookin; HN dev discourse proxy
20. S20 [dev_web] Show HN: Frouter – Live-ping and auto-configure free AI models for coding agents — 1 pts / 0 comments; jyoung105; HN dev discourse proxy
21. S21 [dev_web] ForgeCode: Top open source coding agent in Terminal-Bench 2.0 — 4 pts / 0 comments; gk1; HN dev discourse proxy
22. S22 [dev_web] Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview — 393 pts / 148 comments; GodelNumbering; HN dev discourse proxy
23. S23 [dev_web] Show HN: Amber, a capability-based runtime/compiler for agent benchmarks — 1 pts / 0 comments; _nhynes; HN dev discourse proxy
24. S24 [dev_web] Claude Code ranks 39th on terminal bench. The leaked source shows why — 4 pts / 2 comments; joozio; HN dev discourse proxy
25. S25 [dev_web] Show HN: Wozcode – double Claude Code output — 4 pts / 2 comments; bcollins34; HN dev discourse proxy
26. S26 [dev_web] Show HN: Hm – a task runner with a Python DSL, growing into a CI/CD system — 1 pts / 0 comments; suis_siva; HN dev discourse proxy
27. S27 [dev_web] Show HN: Workplane – collaborative filesystem for humans and AI — 4 pts / 0 comments; tweezers0x; HN dev discourse proxy
28. S28 [dev_web] Codex has dethroned Claude as the king of AI programming — 3 pts / 1 comments; galaxyLogic; HN dev discourse proxy
29. S29 [dev_web] Show HN: GridPath – Faster and Better Agent for Spreadsheets (Tauri, Rust) — 1 pts / 0 comments; pixelmash13; HN dev discourse proxy
30. S30 [dev_web] Show HN: Zorilla – vibe-code a 3D game in the browser — 3 pts / 2 comments; algera; HN dev discourse proxy
31. S31 [github] openai/codex — 86279 stars / 12616 forks / 5237 issues; openai; Repo/adoption/build signal
32. S32 [github] pproenca/agent-tui — 93 stars / 11 forks / 6 issues; pproenca; Repo/adoption/build signal
33. S33 [github] rocketride-org/rocketride-server — 3431 stars / 1062 forks / 130 issues; rocketride-org; Repo/adoption/build signal
34. S34 [github] lightdash/lightdash — 5854 stars / 725 forks / 1687 issues; lightdash; Repo/adoption/build signal
35. S35 [github] can1357/oh-my-pi — 7791 stars / 629 forks / 217 issues; can1357; Repo/adoption/build signal
36. S36 [github] stevesolun/ctx — 371 stars / 47 forks / 1 issues; stevesolun; Repo/adoption/build signal
37. S37 [github] microsoft/skills — 2400 stars / 268 forks / 49 issues; microsoft; Repo/adoption/build signal
38. S38 [github] gastownhall/gascity — 840 stars / 272 forks / 422 issues; gastownhall; Repo/adoption/build signal
39. S39 [github] l3gi0nXXXX/Metis-agent — 130 stars / 13 forks / 0 issues; l3gi0nXXXX; Repo/adoption/build signal
40. S40 [product] Anthropic Claude Code — N/A docs; Anthropic; Product baseline
41. S41 [product] OpenAI Codex CLI — N/A live via GitHub page; OpenAI; Product/repo baseline
42. S42 [benchmark] SWE-bench — N/A benchmark site; SWE-bench; Benchmark baseline
43. S43 [benchmark] Terminal-Bench — N/A benchmark site; Stanford/Terminal-Bench; Benchmark baseline
44. S44 [product] Cursor Agents — N/A docs; Cursor; IDE agent baseline
45. S45 [product] GitHub Copilot coding agent — N/A docs; GitHub; Enterprise coding-agent baseline

9Data Quality / Scan Health

Status: QUALITY_GATE_PARTIAL. Counts: {'dev_web': 30, 'github': 64, 'papers_product': 10, 'reddit': 25, 'youtube': 20, 'x': 0, 'facebook_public': 0}. Gates: source_volume=True, social_completeness=False, cited_30_possible=True. Caveat: papers_product=0 do arXiv 429/timeout; Facebook public=0 usable; X dùng search fallback, direct unauth N/A → confidence Medium.