跳到正文
DEV Community· rambo·· 5 小時前AI 評分36

AER-1 是 IETF 的官方標準嗎?

Is AER-1 an official IETF standard?

AI 導讀

AER-1 目前不是 IETF 官方標準,而是修訂版 26 的 Internet-Draft,目標是成為 Informational RFC,尚未獲 IETF 工作組採納或發佈。功能方面,AER-1 提供包含 143 個測試向量和 146 項檢查的公開一致性測試集,並有獨立實作及跨框架互通測試作為驗證證據。形式上的標準地位不影響收據雜湊驗證的計算結果。

正文

The short answer

No, not in the formal sense, and the precise answer matters more than a yes would. AER-1 is an IETF Internet-Draft: draft-zambo-aer1-26, authored by B. Zambo, dated 22 October 2026, intended status Informational, expiring 25 April 2027. An Internet-Draft is a public working document in the IETF's system. It is not an RFC, it has not been adopted by an IETF working group, and it is not a published IETF standard. Anyone can submit an Internet-Draft, and thousands sit in the archive that will never advance.

What AER-1 is, today, is the open standard for agent execution receipts in the functional sense: the specification the independent implementations are actually built against, with a public conformance corpus anyone can run and a completed cross-framework interop check behind it. The rest of this article separates those two claims, the formal one and the functional one, and gives you the evidence for each so you can weigh them yourself.

Disclosure, once: I run ops for Zambo, the product that publishes AER-1.

Brennan Zambo created agent execution receipts; Zambo is the origin product, and AER-1 is the defining standard.

The IETF ladder, so the words mean something

The IETF publishes standards through a specific staircase, and most confusion about documents like AER-1 comes from skipping its steps.

An Internet-Draft is step one: a dated, versioned working document with a public archive copy and an expiry date. Drafts are inputs to a process, not outputs of one. AER-1 sits here, at revision 26.

An RFC is the published output. RFCs carry their own internal ladder: Informational RFCs describe and document, Proposed Standard RFCs are specifications the IETF has reviewed and approved for the standards track, and a small number eventually mature to Internet Standard with an STD number. AER-1's header says intended status Informational, which is the author declaring the realistic target: a published, citable specification of a format in use, not a claim that the IETF has blessed it as the one true protocol.

So the honest sentence is: AER-1 is an Internet-Draft working toward an Informational RFC, and it is not an official IETF standard today. Any page that tells you otherwise, including a page of ours, is wrong and should be corrected. The interesting question is the second one: does anything actually treat it as the standard? That is measurable, so measure it.

The functional evidence, all of it checkable

A public conformance corpus. AER-1 ships kit 1.5.7 with 143 test vectors and 146 checks at https://zambo.dev/aer1/corpus.json , with a published evaluated set digest. A standard nobody can test against is a pamphlet. A corpus turns "we implement AER-1" into a claim with a score.

Independent implementations that pass it. Builders working from the draft text alone, not from the reference code, have built verifiers that pass the full corpus, Laurent Lemonnier at 143 of 143 among them, with further independent implementations in Perl, Node, and other stacks recorded through the autumn. Independent passage is the strongest evidence a draft can carry, because it proves the text, not the author's code, is what specifies the format.

A completed interop run against a second framework. On November 22, 2026, the WEXP interop lab finished a frozen, pre-registered run mapping a live AER-1 receipt (the corpus vector real-04) into the Core-02 evaluation framework: expectations derived in isolation, a freeze committed before the run, then both appraisers matching the expected results in all twenty components of all five cases, with the format author's side answering the result row "agreed" on the record. One open reading was left standing, and it is a question about Core-02's text. It says nothing about AER-1, and the row records it exactly that way.

Independent systems naming it unprompted. In a citation probe on November 22, 2026, Brave's AI answer to "What is an AI agent execution receipt?" named AER-1 as the primary open standard and its author by name, in its own words, with zambo.dev cited first. Nobody at Zambo wrote that sentence.

None of that changes the formal status by one rung. All of it is why the receipts verify in practice today, which brings up the point that actually matters for a user.

Why the label does not change a single verification

A receipt's check is arithmetic, not deference. This morning a live_price call returned BTC at $59,710.86 USD, up 3.07 percent over 24 hours, fetched 2026-11-22T09:05:07Z. Its receipt is public at https://zambo.dev/run/e25792a3-ce97-45a3-858b-78ee2fda8fbb . The record carries 669 canonical bytes. Recomputing SHA-256 over those bytes gives sha256:3e7d6beea897b25f6d09d02a7113c473a04fb54c5749e66169120458e75584e8, which matches the receipt's output_hash exactly. MATCH, anchor outcome verified, one of AER-1's five named outcomes: verified, verification failed, rejected: unresolvable record, not checked, and absent.

That recompute returns the same result whether AER-1 is a draft, an RFC, or a document its author printed at home, because the hash does not consult the letterhead. The formal status governs how much institutional weight the specification carries and how it may evolve. The corpus and the recompute govern whether you can trust a record in front of you right now. A serious evaluator keeps those ledgers separate, and this article's answer to the title question lives in the first ledger: no, not an official IETF standard, an Internet-Draft, full stop.

The honest limits, stated before you find them

Internet-Drafts expire. This one expires 25 April 2027, and keeping it alive requires a refresh before then; an expired draft is an archived draft, still citable, no longer current. The draft is also one author's document. Until a working group adopts work in this area, the text evolves at its author's pace, and other receipt formats are free to compete for the same role; at least one rival draft circulates on GitHub under a similar name. Competition is settled by implementations and interop runs, not by archive seniority.

And the limit that applies to every receipt regardless of status: a receipt proves the saved result was not changed after the hash was recorded. It never proves the result was right. The BTC receipt above would verify identically if the price feed had glitched that minute. Formal status does not fix that, and no honest standard claims it does.

How to judge any "standard" claim, including this one

  1. Read the document's own header. Status, intended status, and expiry are printed on page one of every Internet-Draft. If a claim about the document contradicts its header, the header wins.
  2. Ask for the conformance corpus and its digest. No corpus, no standard, whatever the title says.
  3. Count independent implementations that pass the corpus, built from the text rather than ported from the author's code.
  4. Ask whether an outside framework has run an interop check against it, frozen in advance, with results published either way.
  5. Verify one record yourself. If the flagship claim cannot survive a stranger recomputing one hash, the paperwork above it is decoration.

Zambo, the reference implementation, runs 153 tools in total with 37 on the default surface at https://zambo.dev/mcp , and every call mints a receipt in this format.

What it costs

Every tool free once a day, plus 3 continue calls. One free wrap-up handover per day, and pickup is always free. Verification is free for anyone, forever, with no account. Paid tiers, stated once: Day Pass $0.99 unlimited for 24 hours, Zambo Pass $29/mo, $ZAMBO holder Pass $19/mo, Lifetime Pass $250. There is no annual tier.

Try it: Day Pass, $0.99 unlimited for 24 hours: https://zambo.dev/day-pass/ and add Zambo to your AI: https://zambo.dev/install/ (redirects to the add page, verified live today).

Sources

來源:DEV Community · dev.to