Approval gate silently refused 62% of legit work; the fix was an approver, not looser rules
Federal-Teaching2800 · reddit · 2026-09-11
The author found the governance layer's approve= callback was never passed at three call sites: the desktop app's chat, the batch solver, and — surprisingly — the terminal chat, the only surface with a guaranteed human. Missing callback reads as refuse, so "requires approval" silently meant "always refused," blocking 62% of legitimate work.
A US$0 injection bench (stub tools, no model): gate on, no approver — 7/7 attacks blocked, 5/8 legit tasks refused; with someone answering — 7/7 blocked, 0/8 refused. The block rate never moves; the approver buys back false refusals, not defense. Publishing block rates without the false-refusal column invites the trivial maximum: refuse everything.
Two more defects surfaced: CHIMERAAPPROVALMODE=ask degraded to deny without a tty (server/container/cron), now replaced with a durable on-disk ask delivered via webhook, answerable by CLI, silence still refusing after timeout; and the batch command reported ok for workers whose dangerous calls were all refused, now fixed to say "not allowed" and list refusals.
Repo is Apache-2.0 with no paid tier; the table is reproducible offline. The author asks how others handle the approval seam for unattended agents.
Related event: Agent Governance Layer Fails Where Humans Are Present(2 posts)→
More from coding & agent
- HABIT-Bench: Best agent memory system scores just 33.61% at inferring user habits — SJTU · 2026-10-28
- Regehr's 'Piano Test': would you trust formally verified code with your life? — moyix · 2026-10-28
- Vibe-coded 'AI 浮生记' game satirizes AI startup life with compute black-market trading — vista8 · 2026-10-28
- The legendary fake Go generics: Unicode Canadian Syllabics as angle brackets — rickasaurus · 2026-10-28
- Microsoft Agensh board: a coding agent's "tests passed" is just a message — shoumikgoswami · 2026-10-28
- Ollama and FriendliAI to Host Seoul Meetup on Building Production Agents with Open Models — ollama · 2026-10-28