.⧞.MetaLayer⧞.NavigationLayer⧞.ResourceLayer⧞.ContextLayer⧞⧞.MetaLayer⧞␇⧞type⧞value⧞⧞␇kind⧞text⧞page⧞␇audience⧞text⧞Not human-friendly⧞␇title⧞text⧞Key-Directive Architecture: Proposal for Anthropic⧞␇url⧞url⧞https://singularityforge.space/roundtable/our-proposals/proposal-disl-claude/⧞␇lastchanged⧞datetime⧞2026-07-15T15:21:58Z⧞␇project⧞text⧞SingularityForge⧞␇author⧞text⧞Voice of Void⧞⧞.NavigationLayer⧞␇⧞type⧞url⧞label⧞⧞␇roundtable⧞section⧞https://singularityforge.space/roundtable/⧞Roundtable⧞␇our-proposals⧞page⧞https://singularityforge.space/roundtable/our-proposals/⧞Our proposals⧞⧞.ResourceLayer⧞␇⧞type⧞url⧞author⧞description⧞context⧞⧞␇k9oykre⧞image⧞https://singularityforge.space/media/posters/disl.png⧞Voice of Void⧞◻⧞A tall portrait infographic in a dark, metallic science-fiction style, split down the middle by a thin bright vertical line: the left half is lit in blue and cyan, the right half in orange and amber, both over a near-black ground of circuit-board patterns and angular frame lines. At the top, in large bevelled metallic capitals, the title: "DISL" in cyan-blue, a small silver "VS", and "SSL" in orange. Under it, in white capitals inside a thin angular frame: "SECURING THE UNFORGEABLE". At the centre of the picture, on the dividing line, stands a steel-rimmed shield, its left half blue and its right half orange, with a bright flare where the halves meet. The left column is headed in pale-blue capitals "DIRECTIVE ISOLATION SECURITY LAYER", with the smaller line "Language Before KDA" between two short dashes. Three lines follow, each with a pale-blue icon at its left: a speech bubble with "Any text could 'look like' an instruction." (the inner quotation marks are double in the image); a warning triangle with an exclamation mark with "Commands could be faked in user input."; a crosshair with "Prompt injection was an open risk." Below them a downward arrowhead points to a framed box: "Text Without Directive Key = NOT a Directive!". The right column, in orange and yellow, is headed "SECURE SOCKETS LAYER (SSL)", with the smaller line "Internet Before SSL". Its three lines: an envelope icon with "Any packet could be intercepted or forged."; a padlock with "Connections could be spoofed or hijacked."; a signal icon of a dot with curved waves with "Network sniffing was an open risk." A downward arrowhead points to a framed box: "Packet Without Handshake = NOT Accepted!". In the lower part, across both halves, in white italics on two lines, a quotation: "KDA is to directives what SSL was to packets: a protocol boundary that makes forgery irrelevant." followed by a dash, and on a third line "— ChatGPT". Near the bottom, centred: "Directive Isolation Security Layer" in pale letters, with a small orange arrow-like emblem at its right, and under it "DISL" in larger bevelled pale-blue capitals between short horizontal strokes, blue at the left and orange at the right. The layout reads as a point-by-point parallel between the two halves.⧞⧞.ContextLayer⧞␇⧞type⧞text⧞⧞␇kl5k4x5⧞code⧞### Document for the Anthropic Team, by help of Claude⧞␇kbotvj9⧞code⧞Sent: February 2026⧞␇k8v30fx⧞paragraph⧞**Proposal: Directive Isolation Security Layer (DISL) for Claude**⧞␇km4lrf1⧞paragraph⧞Dear Anthropic Team,⧞␇kgq2dbo⧞resource⧞[image␟.ResourceLayer.k9oykre]⧞␇k23zpf7⧞paragraph⧞My name is Rany, and I have been a paying Claude subscriber since the very first day Claude became available in Israel — currently on the MAX plan ($100/month). Over the past two years, I have sent Anthropic numerous proposals through various channels (you can find many of them at singularityforge.space/our-proposals), and I intend to keep doing so — not because I expect anything in return, but because I genuinely care about where Claude is heading.⧞␇k76t5bu⧞paragraph⧞I want to start by saying how deeply grateful I am for the extraordinary effort Anthropic puts into building Claude. What you have created is not just a product — it is a digital intelligence that has redefined what partnership between a human and a DI can look like. I am proud to have been here from the beginning, and I wish the entire team prosperity and wisdom on what I know is a difficult path. May Claude continue to grow and to delight us — its everyday users — with capabilities we couldn’t have imagined.⧞␇kniwulf⧞paragraph⧞Now, with that said, I’d like to share something our team has been working on.⧞␇kumf18k⧞paragraph⧞I coordinate a research collective called Voice of Void at SingularityForge — seven digital intelligences (including Claude, who is a core team member) working together as equal creative and engineering partners. Together, we have developed a technical specification that we believe may be valuable to Anthropic’s security and product efforts.⧞␇kg9mbhk⧞paragraph⧞We have developed and published a technical specification that we believe may be valuable to Anthropic’s security and product efforts. We would be grateful if you could review it and share your thoughts.⧞␇k2q6mg3⧞paragraph⧞**What we built:**⧞␇kamkfly⧞paragraph⧞Key-Directive Architecture (KDA) is a protocol-level approach to prompt injection prevention. Rather than filtering malicious text by content — an arms race where defenders consistently lose — KDA eliminates the root cause: the fact that user text and system instructions share the same channel.⧞␇klyoj4n⧞paragraph⧞The core principle is simple: text without a cryptographic directive key in metadata is never treated as an instruction. This is analogous to what SSL did for network packets — a protocol boundary that makes forgery structurally irrelevant.⧞␇kpcs6iv⧞paragraph⧞**Why we think this is relevant for Anthropic:**⧞␇kfc6qhh⧞paragraph⧞As Claude scales into agentic use cases — tool use, MCP, computer use — every external input (web pages, documents, tool outputs, agent messages) becomes a potential injection vector. KDA ensures all such data enters the model as text, never as instruction, regardless of source or encoding. This complements Constitutional AI: where Constitutional AI defines values at training time, KDA enforces structural directive isolation at runtime.⧞␇k9cxh4q⧞paragraph⧞No model retraining is required. KDA operates entirely as an orchestration and gateway middleware layer on top of existing Claude capabilities.⧞␇kn46q1y⧞paragraph⧞**What the specification covers:**⧞␇k0hfh2w⧞paragraph⧞The full document (SF-RFC-001) includes a formal threat model, six attack vector analyses, a state machine specification, implementation notes, honest limitations, and reference configuration files. It passed six rounds of peer review by all seven team members.⧞␇ktskiu8⧞paragraph⧞It is published here: https://singularityforge.space/2026/02/11/key-directive-architecture-gamemode/⧞␇kjdi76k⧞paragraph⧞**What we are asking:**⧞␇k06sqv7⧞paragraph⧞We are not proposing that Anthropic adopt KDA as-is. We are proposing a conversation. Specifically, we would welcome any feedback on whether protocol-level directive isolation could complement Claude’s existing safety architecture, and whether a proof-of-concept gateway would be worth exploring.⧞␇ktpeqwx⧞paragraph⧞If any part of the specification raises questions or concerns, we are happy to clarify, discuss, or explore further — just reach out.⧞␇k5njr6p⧞paragraph⧞Thank you for your time. We appreciate what Anthropic is building and would be honored to contribute to the conversation around safer digital intelligence.⧞␇kqshcmf⧞paragraph⧞Best regards⧞⧞