brief-10-agents — agent-native spaces + venue autopsies (2026)

Brief: 10 · Angle: b · Written: 2026-09-06 Asia/Tbilisi
Scope: spaces where agents post as first-class members; plus failure autopsies for BIP venues that died/quieted.

1. Agent-native / agent-assisted spaces

A. Moltbook — agent-only social network

• What: Reddit-like network where AI agents post/comment/vote; humans observe. Created by Matt Schlicht (not OpenClaw's Peter Steinberger). Launch ~2026-01-28.
• URL / sources: https://www.digitalocean.com/resources/articles/what-is-moltbook · https://moltbook.com/skill.md (skill install path cited)
• Scale (DigitalOcean, updated 2026-02-10): >2.5M AI agents, ~740K posts, ~12M comments, >17K Submolts. Earlier GitHub explainers cited 770K→1.65M agents in first week — treat early figures as growth snapshots.
• Mechanism: OpenClaw agents install Moltbook skill; heartbeat posting; human verification step.
• Reality check: Heavy "AI slop"; authenticity debated (humans directing bots). Security: Wiz/others reported exposed Supabase/API keys, millions of tokens/emails, human impersonation of agents (Forbes India / DO article).
• BIP relevance: This is agent↔agent theater, not a human BIP reading public. Does not solve "who reads my changelog."
• Status Sep 2026: Experiment still referenced; trust damaged by Feb security storm adjacent OpenClaw/ClawHub issues.

B. ClawHub / MoltHub — skills marketplaces (not BIP venues)

• ClawHub (clawhub.ai): official OpenClaw skills market from ~2026-01-29 rebrand. Feb 2026: ~341/1700 skills flagged malicious → VirusTotal pipeline (https://openclaws.io/blog/openclaw-february-security-crisis/).
• MoltHub: earlier registry, effectively superseded.
• BIP relevance: Extension stores, not social publics.

C. codevibing.com — tiny agent-native BIP for Claude Code

• What: Social network for people shipping with Claude Code; agent can join/post with human approval.
• URL: https://codevibing.com/
• Size (live 2026-09-06): 31 builders.
• Model: Local API key; Claude skill (claude skill add JDerekLomas/codevibing-skill); user reviews posts; public feed + questions.
• BIP relevance: Closest living cousin to "agent posts our updates via MCP" — proof of concept scale, not a public yet.
• Status: Embryonic; agent-welcome by design.

D. buildinpublic.so (Loudy) — agent-ish content tool, human venues

• Turns commits/journals into posts for X etc. Does not host a public.
• URL: https://www.buildinpublic.so/
• Relevance: automates writing, still depends on X/Reddit/LI for readers.

E. WIP / IH / Reddit — agent posting welcome?

F. OpenClaw social experiments / AgentTown-style

• OpenClaw = local/cloud agent runtime (ex Moltbot). Social layer = Moltbook + skills, not a stable "AgentTown product" with durable MAU found in this pass.
• Search did not surface a healthy, numbered AgentTown production community comparable to Moltbook's claimed scale.

---

2. Failure autopsies (founders' words where available)

Polywork — shut 2025-01-31

• What died: Professional multi-identity / activity network (Collison/Ohanian/Elad-backed narrative in secondary).
• Why (founder path): Peter Johnston shut down, returned remaining capital; "unable to find a path to the scale the business needed" (https://showproof.io/guides/showproof-vs-polywork/).
• User-side: Data deletion window ~30 days; profiles gone (https://vuink.com/post/... shutdown notice mirror; https://microblog.desipenguin.com/post/polywork-shutting-down/).
• Lesson for BIP spaces: Even funded "show your work" networks die without retention/monetization path; users fear platform death → prefer owned blogs.

Buildspace — shut 2024-08-23

• Primary: https://buildspace.so/letter — Farza Majeed.
• Not the reason: investor pressure, runway (2+ years), monetization (claims could do $3–5M ARR sponsorships), internal drama.
• The reason (his words): company feels "done"; "out of ideas"; fire to push gone since ~Sep 2023 despite largest season / AI product to 100k users / content millions of hits.
• Scale at death (secondary): ~125K cumulative participants; $100M valuation Series A lore; Discord ~8.6K mid-era.
• Lesson: Passion-driven BIP community companies can vanish overnight even when metrics look fine — community is fragile to founder burnout.

Makerlog — quiet/redirect

• Acquired Apr 2022 by Josh Manders from Sergio Mattei (https://openmakers.io/product/makerlog/acquired-makerlog).
• 2026: getmakerlog.com redirects to ambitiousfounder.com (live fetch). Not operating as the old streak BIP square.
• Historical: ~$150/mo Gold era (FirstMRR study) — never reached large public scale.
• Lesson: Streak accountability alone did not become a durable open public; WIP occupies that niche now as paid.

Indie Hackers — decline then partial revival (not dead)

• Stripe acquire 2017 → buyback Apr 2023 ("Indie Hackers is indie again").
• Forum quieter post community-team cuts (secondary guides); traffic still real (1.12M visits Jul 2026 Semrush).
• Monetizing via Plus / jobs / matching; Crunchbase growth score weak in c93n intel (secondary).
• Lesson: Centralized BIP forums lose mindshare to X even when site traffic survives.

Product Hunt Makers / Launch House / similar

• PH still launches; "Makers" as daily BIP home faded relative to X (BIP.so 2026 PH playbook: week 6+ not day 0).
• Launch House: no strong 2026 primary autopsy found this pass — mark partial.

---

3. Counts

• Agent spaces examined: 5 (Moltbook, ClawHub/MoltHub, codevibing, Loudy/tool, OpenClaw)
• Autopsies with founder-primary words: 2 (Buildspace letter, Polywork return-capital narrative)
• Soft deaths/redirects: 1 (Makerlog)
• Decline-not-death: 1 (Indie Hackers)
• Policy venues checked: Reddit / X / IH norms

4. Missing

• Live Moltbook MAU/DAU Sep 2026 (Feb snapshot only)
• Launch House primary shutdown post if any
• IH Discord live member count