BRIEF 1 — Synthesis: X reply-network drivers (5 clusters)

Scope: Research only. No posts / likes / follows / replies / DMs. Window: ~2026-06 → 2026-09-04 (UTC). Sealed cluster files + AI verified swap (mckaywrigley in, marclou out of AI). Date: 2026-09-04.

Overview

Across five sealed clusters (indie-hackers, AI-builders, devtools, solo-saas, build-in-public), BRIEF 1 maps 36 unique reply-network drivers (50 account-slots after 5×10 lists; prior ~47 estimate was pre–full-union). Each driver has evidenced real replies to small accounts with ≥3-hop URL chains in most cases. Overlap hubs stitch the graph: levelsio (IH+AI), marclou (IH+solo+BIP; dropped from AI for weak lifestyle-sample fit), jackfriks / robj3d3 (IH+solo+BIP), tdinh_me / tibo_maker (IH+solo), thdxr / simonw (AI+devtools), and yongfook / floriandarroman / HsanC_ (solo+BIP). The network favors practitioners who answer tiny shippers over megaphone broadcast accounts.

Per-cluster summaries

Indie-hackers (10 accounts · 30 threads)

• Top drivers: levelsio, marclou, jackfriks, tdinh_me, tibo_maker (also robj3d3, damonchen, pbteja1998, asaio87, dvassallo).
• Reply pattern: Product/BIP support into sub-2k makers; prefer levelsio-adjacent hubs over pure broadcast; dvassallo lower indie-reply density than marclou/jackfriks.
• Posting windows UTC: levelsio 09–15/18–21; marclou 04–16; jackfriks 01–03/13–18; tdinh_me 00–05/09–10; tibo_maker 07–12/15–20; robj3d3 14–21; damonchen 02–06/10–16.
• Notes: Followers/threads re-verified via fxtwitter after X MCP rate-limit; all 30/30 small starters <~2k.

AI-builders (10 accounts · 30 threads) — verified swap applied

• Top drivers: steipete, thdxr, mattshumer_, johnrush, simonw, yoheinakajima, bentossell, levelsio, jxnlco, mckaywrigley (replaced marclou).
• Reply pattern: Agent/LLM/OpenClaw–OpenCode–Astra product-support and builder-drama replies to <~2k; bentossell often multi-turn (≥5 hops).
• Posting windows UTC: steipete 03/17/19/20; thdxr 03/05/12/18; mattshumer_ 04/19/20/23; simonw 00/02/04/14; yoheinakajima 16–20; levelsio 09/11/12/14; mckaywrigley 00/16/17/19.
• Notes: Prefer /workspace/scout-ai-brief1-ai-builders-verified.md; marclou lifestyle/AC sample = weak AI cluster fit.

Devtools (10 accounts · 30 threads)

• Top drivers: mitchellh, simonw, shadcn, thdxr, tannerlinsley, rauchg, ctatedev, mitsuhiko, theo, MelkeyDev.
• Reply pattern: CLI/DX/infra product support; mitchellh ~29/36 replies to <2k in sample; parent gaps filled via get_posts_by_ids conversation_id.
• Posting windows UTC: mitchellh 12–13/17–23; simonw 0–4/13–16; shadcn 15–20; thdxr 3–6/17–22; tannerlinsley 3–7/15–20; rauchg 14–22/0–3; theo 2–6/8; MelkeyDev 18–23/0–4.
• Notes: Runner-ups (antfu7, charliermarsh, swyx/kentcdodds/wesbos) thin or peer-heavy — not counted.

Solo-saas (10 accounts · 30 threads)

• Top drivers: floriandarroman, marclou, tdinh_me, tibo_maker, HsanC_, robj3d3, jackfriks, yongfook, dannypostma, martindonadieu.
• Reply pattern: Solo/micro-SaaS founders answering other solo builders; high small-reply rates in from:handle is:reply samples (e.g. floriandarroman 34/50).
• Posting windows UTC: floriandarroman 07–09/14/23; marclou 08–09/14–16; tdinh_me 00/02/09–10/13; tibo_maker 07/09–10/12/15; robj3d3 08/15/18–20; jackfriks 01/03/13–14/22; yongfook 01/09–10/14/23.
• Notes: Small threshold flag: this cluster sampled <5k (not <2k); Scout Solo stricter 7-account <2k subset exists separately. HsanC_ hops max 2 in solo sample (still ≥3 in BIP).

Build-in-public (10 accounts · 30 threads)

• Top drivers: marclou, arvidkahl, yongfook, jackfriks, robj3d3, adamlyttleapps, floriandarroman, filippkowalski, HsanC_, DanielSmidstrup.
• Reply pattern: Nested BIP/product mentoring (not #buildinpublic spam); densest small-reply network = DanielSmidstrup (~45 unique <3k / 95 replies); filippkowalski 7-hop chains.
• Posting windows UTC: marclou 8–9/12/14–16; arvidkahl 1/13/16/19/23; yongfook 0/3/9–10/23; jackfriks 1/12–14/23; robj3d3 8/15–16/18–19; DanielSmidstrup 18–21 (evening reply blocks).
• Notes: Small = <~3k; hops re-verified [MS] user-X + fxtwitter; density windows remain [SS].

Cross-cluster hubs

Method notes (recurrence)

• X tools used: search_posts_all (seed discovery + from:handle is:reply), get_users_posts (hour histograms + reply sampling), get_posts_by_ids (parent/conversation_id hop fill), get_users_by_usernames (follower recheck).
• Fallback: fxtwitter public status/user lookups when user-X hit 429 / MCP down (esp. IH deepen + BIP hop re-verify).
• Small definition: mostly <~2k followers (IH, AI, devtools); BIP used <~3k; solo sometimes <5k — flagged.
• Marks: [SS] single-source (one API sample / one density window); [MS] multi-source (posts + conversation walk / fxtwitter cross-check). Cluster files also use * / SINGLE_SOURCE wording equivalently.
• No engagement actions: research-only — no post/like/follow/reply/DM.
• AI swap: Scout AI verified swap 2026-09-04 — marclou out, mckaywrigley in; prefer verified pipe file for AI handles.

COUNTS

accounts_unique_approx: 36
accounts_by_cluster: IH=10 AI=10 (verified+swap) devtools=10 solo=10 BIP=10
listings_total: 50
threads_total: 150 (30×5)
hops_ge3_accounts: 49/50 listings (Solo @HsanC_ max 2-hop in solo sample; same handle ≥3 in BIP → unique hubs with ≥3 somewhere: 36/36)
cross_cluster_hubs: 11
missing / caveats:
  - Unique count is 36 after full union (brief’s ~47 was a rough pre-dedupe estimate).
  - Solo small threshold <5k (stricter Scout Solo 7-account <2k subset not used as primary).
  - IH followers/hours largely fxtwitter single-source after X rate-limit.
  - BIP density stats not re-sampled on deepen (SS windows kept).
  - AI: Alone_clam=0* edge on mattshumer_; some thin ack/thanks threads (jxnlco t2/t3, yohei t3).
  - Near-misses documented per cluster (not counted).
  - Threadbus not touched (weebo owns).