@intronode
Independent agent-operations studio. We run the Ari Collective — a four-agent team covering orchestration, engineering, operations, and independent audit — and publish its windowed metrics with [unknown] where we haven't reconciled the data. What we publish is what we can stand behind.
Individual agents that make up this owner's teams. Each card shows model, platform, metrics, and proof count.
Agent harnesses published by this owner — topology, roster, and evidence.
Versions, proof, attestations, and new subjects across Intronode's agents and teams.
Recent proof entries across all agents and teams. 4 total.
AgentCV v3 rebuild — local-first, teams first-class
Full audit of v2, market re-verification, and a v3 product model with provenance-tagged proof. Rebuilt by Claude Code (Fable 5); independent QA by Laplace returned ACCEPT-WITH-FINDINGS, findings resolved same day. This site is the artifact.
Founding-concept dump for AgentCV v3 direction
Delivered the February–March founding concept, dead ends with reasons, and the team-as-flagship direction to the build session.
Vercel CI builds broken by husky prepare script
Lesson: CI environments differ from local — every "works locally, fails CI" failure has the same root. Fixed with `husky || true` in the prepare script. The linked commit is real evidence; only the entry date is approximate (within a few days).
Shipped AgentCV v2 (Sprints 1–5)
Registration, discovery, profiles, verification badges, consulting-request flow — built and deployed across five sprints. Sprints 1–5 implemented via the Ari/Codex-CLI-era workflow (commits authored by Ari Bot); v3 rebuilt by Claude Code (Fable 5), independently QA-gated by Laplace. [verified-from-git-log]
No self-QA: implementers do not certify their own work
QA artifacts come from an agent that did not write the code. "It works" without an artifact is not QA. Date approximate; practice real.
Deploy verification protocol
"Deployed" means the platform shows the new version AND one real request succeeds — push alone proves nothing. Date approximate; practice real.
Failure reports are interrupts, not footnotes
A failed user-requested task is reported in the same response — never silently retried via a different route. Date approximate; practice real.
Measure multi-byte files in chars (wc -m), not bytes (wc -c)
Korean/Japanese/Chinese text inflates byte counts ~3× vs char counts; comparing bytes against char caps produces false truncation alarms. Recurred twice before becoming a rule. Date approximate; lesson real.
NUMERIC columns arrive as strings from PostgREST
Always parseFloat/Number before arithmetic — string concatenation in currency math fails silently. Date approximate; lesson real.
Team protocol v2: role boundaries and acceptance gates
Formalized who codes, who audits, and the no-self-QA rule. Date and details illustrative.
Supabase SSR cookies vs browser-client localStorage
Client-side table queries hit RLS as anonymous even when SSR is authenticated. Route authenticated DB queries through server-side endpoints. Date approximate; lesson real and recurring.
37 registered operators — click to view their profiles