chore(openclaw): full agent architecture review — post-Reading-B S3 fractal alignment

[Argocyte]

Full review of the OpenClaw agent roster against the current S3 fractal model. Current agents (Clerk, Steward, Sentry, Librarian, Orchestrator stub) predate Reading B, sociocratic self-similarity, MVM, and the Bonfire/AP federation direction. Review assesses: which agents remain, which need S3 domain-circle reframing, which need sub-domains, how the historian domain fits, how missing roles (DPO, Education Officer, etc.) map. Consented AGM 2026-04-12 (D10). Design plan: d10-openclaw-agent-review.md. Filed by Et under G5.

[Argocyte]

Update from Iskander-Code development (2026-04-12)

The agent architecture review (D10) has been resolved in Iskander-Code. The old domain-named agents (clerk, steward, sentry, orchestrator, historian, researcher, archivist, communications) have been replaced with 7 principle-named agents, each IS one ICA principle:

Agent	Principle
membership	P1: Voluntary & Open
democratic-control	P2: Democratic Control
economic-participation	P3: Economic Participation
autonomy	P4: Autonomy
education	P5: Education
cooperation	P6: Cooperation
community	P7: Community

The facilitator is a deterministic algorithm (not AI) that mediates between agents. S3 all the way down — each domain can spawn ets own subdomain with ets own facilitator.

Values Council (10 guardians) serves as supreme court for Et+human deadlocks.

23 skills in knowledge commons. 90 tests, zero failures.

See: `Argocyte/Iskander-Persistent-Sociocratic-AI-Architecture`