Resources

Resources: doctrine

The doctrinal corpus: white paper, 4 principles, 4 delegated autonomy levels, DDAO role, glossary, toolkit manual (14 parts) and deck. The skeleton of the ACF® standard.

iNote

Doctrinal resources are the root of everything: the cards reference them, the regulator guides map onto them, and REASON tools invoke them in their rationale. If something stays unclear, you are likely reading the wrong layer, come back here.

What this contains

27 resources: 9 structured JSON bricks (4 principles, 4 delegated autonomy levels, the DDAO), the white paper, the deck, and the full manual with its table of contents and its 14 individually addressable parts.

Exhaustive list

Root documents

acf://whitepaperMarkdownFR + EN

ACF® White paper

The ACF® white paper, same text as the edition published on acfstandard.com.

acf://framework/ddaoJSONFR + EN

DDAO · Delegated Decision Agent Officer

Definition of the ACF® role that carries accountability for delegated decisions, independent from the team that owns the agent.

acf://deckMarkdownFR + EN

ACF® Slide deck

Introduction deck of the standard (27 slides), extracted from the version published on acfstandard.com.

Founding principles ×4

See also: READ tool acf.search to search a principle by keyword.

acf://framework/principle/P1JSONFR + EN

P1 · Decision Sovereignty

Critical strategic decisions are never delegated to autonomous agents. The organisation keeps ultimate accountability for the decisions its agents take, whatever level of autonomy it delegates to them.

acf://framework/principle/P2JSONFR + EN

P2 · Governance by Design

Governance frameworks are defined before deployment, not after, and their intensity is proportionate to the criticality of the decisions entrusted to the agent.

acf://framework/principle/P3JSONFR + EN

P3 · Ultimate Human Control

Every agentic system must preserve the ability for human intervention: a takeover mechanism that can be activated at any time and is calibrated to criticality.

acf://framework/principle/P4JSONFR + EN

P4 · Traceable Accountability

Every autonomous action must be auditable and attributable: any decision can be reconstructed after the fact by pointing to the doctrine, rules and data that produced it, and to the person who authorised it.

Delegated autonomy levels ×4

See also: REASON tool acf.assess-autonomy for an agent to recommend an autonomy level for a given decision.

acf://framework/autonomy-level/suggestiveJSONFR + EN

Suggestive

For a given decision, the agent suggests and the human decides. The human keeps hold of every executing action.

acf://framework/autonomy-level/co_decisionJSONFR + EN

Co-decision

For a given decision, the agent proposes a reasoned option and the human validates or rejects it. Every executing action goes through human validation, on an already framed option.

acf://framework/autonomy-level/supervised_autonomousJSONFR + EN

Supervised autonomous

For a given decision, the agent decides and executes within a bounded perimeter, under supervision. Beyond the thresholds defined in the mandate, it must escalate the decision to a human.

acf://framework/autonomy-level/autonomousJSONFR + EN

Autonomous

For a given decision, the agent decides and executes without prior human validation, within the mandate's perimeter. Human control remains possible after the fact and the stop mechanism remains active.

Pedagogical manual × 1 + TOC + 14 sections

The manual is served in French and English, extracted from the files published on acfstandard.com.

acf://manualMarkdownFR + EN

ACF® Toolkit manual, full

The full toolkit manual (331 pages) in one payload. Large: prefer the table of contents and the sections.

acf://manual/tocJSONFR + EN

ACF® Toolkit manual, table of contents

The 14 parts of the manual with their titles, page ranges and section URIs.

acf://manual/section/{1..14}MarkdownFR + EN

ACF® Toolkit manual, one part

One of the 14 parts of the manual. Stable URI, useful to quote a precise passage.

Fetching a doctrine resource

Framework resources return structured JSON (id, code, title, summary, doctrine, related_fiches fields). Narrative resources return Markdown with a served_locale and is_fallback frontmatter at the top.

fetch-principle.tsts

// Read principle P3: Ultimate Human Control
const { contents } = await client.readResource({
  uri: "acf://framework/principle/P3",
});

const payload = JSON.parse(contents[0].text);
// {
//   "id": "P3",
//   "code": "P3",
//   "title": { "fr": "Contrôle humain ultime", "en": "Ultimate Human Control" },
//   "summary": { "fr": "...", "en": "..." },
//   "doctrine": { "fr": "...", "en": "..." },
//   "related_fiches": ["ACF-07", "ACF-P6", "ACF-P7"]
// }

// Read the whitepaper in French
const wp = await client.readResource({ uri: "acf://whitepaper?lang=fr" });
console.log(wp.contents[0].mimeType); // "text/markdown"
console.log(wp.contents[0].text);     // ---\nserved_locale: fr\nis_fallback: false\n---\n...