Verification-first execution control for high-consequence systems.
"If it cannot be verified, it cannot execute."
Status: Draft; not yet a final standard or certification program.
This version incorporates findings from two independent structured-review memoranda. Publication does not imply endorsement, certification, or implementation conformance by any reviewer or outside organization.
- Current Version: v1.0 — Public Review Draft (second-pass corrected)
- Published: July 2026 (public review draft)
- Publisher: Remnant Fieldworks Inc.
- Built on: RF-000 (Proof Before Power™ Doctrine, published)
- Certification: None exists yet. RF-101 (assessment) and RF-102 (certification) are under development. No party may claim RF-100 certification or formal conformance at this maturity level. This draft is not final, frozen, certified, or approved by any external organization.
Review history:
- v0.9 (July 15, 2026): public draft
- v1.0 unified draft (July 15, 2026): integrated structured peer review findings G-01 through G-18 (James A. Bex, BxTech Systems Solutions LLC)
- v1.0 second-pass corrected (July 16, 2026): integrated second-pass findings F-01 through F-06
RF-100 defines the minimum normative structure a high-consequence system — AI agent, automated workflow, financial execution system, infrastructure controller — MUST satisfy before a governed action executes.
The standard specifies requirements for governing consequential actions ("Governed Actions") proposed by human or AI Actors, such that:
- No Governed Action takes effect without independent verification of authority, evidence, policy compliance, and system state
- Every decision is durably and verifiably recorded
- The proposing Actor cannot modify the verification gate, policy, evidence, or decision
Core elements (Sections 1-13):
- Policy and Execution Boundary Profile (EBP)
- Evidence
- Verification and Decision Inputs
- Authority
- Decisions: Rendering, Validity, and Failure Posture
- Verification Decision Records (VDRs)
- Execution Integrity
- Auditability and Reconstruction
- Enforcement Architecture (E1/E2/E3)
- Gate Governance
- Trusted Time
AI Actor Profile (RF-100:AI) defines elevated requirements for AI agents and autonomous systems, including:
- Agent identity (model/version/lineage in every VDR)
- Scoped capability and authority
- Intent Records (untrusted context, never Evidence)
- Consequence thresholds and velocity limits
- Prompt/input-injection detection
- Chained-agent per-hop governance
- Model/prompt-update re-versioning and re-evaluation
- Current:
RF-100-v1.0-draft.pdf(markdown) — v1.0 unified draft, second-pass corrected - IPR commitment (draft):
IPR-COMMITMENT.md— intended patent posture (royalty-free, scoped, reciprocal); draft pending counsel review - Prior version:
RF-100-v0.9-draft.pdf— v0.9 public draft - Archived / citable versions: Zenodo concept DOI 10.5281/zenodo.21366341 (stable link across all versions)
This draft is open for public review. The draft is strongest where it gets attacked.
- Open a GitHub Issue in this repository for any critique, ambiguity, missing requirement, conflict with existing standards (NIST AI RMF, ISO 42001, SOC 2, etc.), or implementation concern.
- Please reference requirement numbers (e.g.,
RF-100.7.2,RF-100:AI.3) where possible. - Substantive feedback will be credited in the v1.0 revision history unless you request otherwise.
- Not a certification program (yet)
- Not legal or compliance advice
- Not a claim that any deployed system currently conforms
- ExecutionProof™ does not yet claim full RF-100 conformance. Asymmetric (8.4-conformant) signing and other required controls are not yet operational; a conformance claim will follow only when they are.
- No reviewer or outside organization has endorsed, certified, or approved this draft. Independent review findings were incorporated; that is not endorsement.
© 2026 Remnant Fieldworks Inc. The text of the RF-100 specification is released under CC BY 4.0, effective with this v1.0 release (see Section 0.1). This license applies only to the specification text and grants no rights to Remnant Fieldworks trademarks, service marks, certification marks, software, hosted services, APIs, trade secrets, or patents — all of which are reserved.
Remnant Fieldworks™, Proof Before Power™, Verification Before Execution™, ExecutionProof™, VaultProof™, and ProofRecord™ are trademarks or pending trademarks of Remnant Fieldworks Inc. U.S. patent applications pending.
Patent / IPR commitment (draft). A draft IPR commitment describing the
intended patent posture — a royalty-free, scoped, reciprocal (defensive)
non-assertion covenant for conforming implementations, with all rights
reserved for trademarks, products, hosted services, reference code, and the
certification program — is published as
IPR-COMMITMENT.md. It is a draft pending legal /
patent counsel review; the binding version will be finalized before any
certification program (RF-101 / RF-102) launches. In short: anyone may
implement RF-100 freely; no one may copy the ExecutionProof/VaultProof
products, brands, or certification authority.
Contact: derek@ownerremnantfieldworks.com