# kaal:position:2026-09-12-003

**Affirmed position.** Cao calls for tooling that traces an agent's reasoning chain. Kaal's position is that governance needs evidence of effects, not testimony: files, network connections, credentials, and devices observed from outside the agent and tied to the authorization each run used. Without such a record, a user can hold every byte locally and still have no answer to what happened. An unobserved channel is labeled unobserved, never shown as safe.

**Status.** affirmed  **Published.** 2026-09-12

**Holds when.**

- Cao's paper is a position paper citing benchmark evidence, not an empirical test of these claims.
- The position is Kaal's own argument in the essay Prompts Don't Govern Agents. Reputation Does.
- The position does not validate Cao's figures, benchmark methodology, or reported benchmark results.
- The position is limited to observable effects and authorization records for agent runs.

**Current debate.** Zhenfeng Cao, Agentic Software: How AI Agents Are Restructuring the Software Paradigm (arXiv:2606.05608v2, 2026): https://arxiv.org/abs/2606.05608

**Extends.** kaal:claim:7314479-002: https://wulfkaal.github.io/claims/7314479-002

**Scholarly basis.** Wulf A. Kaal, Institutional Requirements for Sovereign Local Agent Runtimes (2026). SSRN: https://ssrn.com/abstract=7314479

**Source PDF sha256.** `debace24a155ae924a155b1fafe98856d98cf83689feff2f87a32f1c06171ce6`

**Topics.** ai-and-agents, institutional-design, risk-and-incentives

**Provenance.** Affirmed in kaalvis:2026-09-12:prompts-reputation-essay-six-v1 at https://wulfkaal.com/2026/09/12/prompts-dont-govern-agents-reputation-does/.

**Record type.** This is a dated commentary position that extends a scholarly corpus claim. It is not a verbatim claim extracted from the paper.

**Canonical form.** This markdown file is the canonical hashed representation of the position.