Spec-driven agent harness for Claude Code and Codex — five-agent pipeline (scope, plan, build, verify, learn) with an auditable proof chain.

Link features, claims, tests, and evidence into verifiable releases stored on Walrus.

Steven Chou's GitHub profile README

Proof History Chain: Convert single signed proofs into continuous, tamper-evident audit history (REMEDA Stage322).

Shared signed-receipt library — DSSE/ECDSA-P256, cosign-compatible, UNSIGNED-honest fallback. The one signing primitive every SZL component shares.

SZL Holdings monorepo — substrate runtime, agentic loops, MCP server (11 tools), reusable workflows, CI gates. Doctrine v11 LOCKED (749/14/163)

SZL Holdings Public Trust Portal — Covenant Proof Standard run artifacts, deterministic replay receipts. Doctrine v11 LOCKED (749/14/163)

Trace-first evaluation harness for deciding whether AI agents deserve more tokens, permissions, and trust

Ouroboros bounded-recursion runtime — agentic loops, dual-witness emitters, governance budgets. Doctrine v11 LOCKED (749/14/163)