Governed pipeline for agent behavior changes · append-only · hash-chained · change/1.0
Flags appear as you type the proposed text.
Proposals land in your local working copy below. Publishing to the canonical log happens via repo PR + Ed25519 seal.
…
Download a signed-bundle-ready JSON record for any change: proposal, full decision trail, and chain verification. Send it to the ledger maintainer to publish via PR + Ed25519 seal.
Advisory review trail, not a sandbox. Risk flags are regex heuristics over added lines — they catch known-bad patterns, not novel ones, and can false-positive on innocent words. The browser working copy is yours alone until a maintainer seals it into the canonical log. Verify any published ledger with verify.js — never on trust.
Don't take our word for it — re-run the check from your own machine or your own agent runtime.
ledger_tasks: read CWI's governed operational records from your own runtime — same append-only, hash-chained discipline. Zero-trust guide: VERIFY.md.Paste a before/after, get heuristic risk flags, record the decision — then verify any published ledger yourself with the zero-trust checker.