From 281f60b8bbd8fd5f8a4f28b85534f5b4bd5b36d9 Mon Sep 17 00:00:00 2001 From: jason1015-coder Date: Sun, 16 Aug 2026 14:51:16 +0100 Subject: [PATCH 1/4] Update scriptura.md with v1 requirements and tasks Added checklist items for updates and changes needed in scriptura to align with Nano-collective's guidelines and requirements. --- content/collective/whitepapers/scriptura.md | 38 ++++++++++++++------- 1 file changed, 25 insertions(+), 13 deletions(-) diff --git a/content/collective/whitepapers/scriptura.md b/content/collective/whitepapers/scriptura.md index 7bb1c41..bfeec93 100644 --- a/content/collective/whitepapers/scriptura.md +++ b/content/collective/whitepapers/scriptura.md @@ -140,21 +140,33 @@ Questions 1 (naming) and 2 (default provider) were settled during the review win Must exist in v1 and after throughout: -- **MUST update** docs of scriptura to match Nano-collective's brand guidelines -- find some contributors, at least **one more core maintainer** (**jason1015-coder alone not practical to do all work**) by a issue in the transferred repo to Nano-collective -- mainwindow.cpp:888 reads settings straight out of QSettings, **MUST CHANGE** to: +- [ ] **MUST update** docs of scriptura to match Nano-collective's brand guidelines + +- [ ] find some contributors, at least **one more core maintainer** (**jason1015-coder alone not practical to do all work**) by a issue in the transferred repo to Nano-collective + +- [ ] mainwindow.cpp:888 reads settings straight out of QSettings, **MUST CHANGE** to: - OS keychain - encrypt it (personal key) -- Create Rust<---> typescript communication layer **MUST BE IMPLEMENTED** -- USE **Nanocoder as the backend AI layer** instead of current existing , roughly sketched AI layer -- **exclude Nanocoder existing TUI** -- UI must **stay C++/QT** -- all backend must route through **RUST BACKEND LAYER, INCLUDE NANOCODER-AI PARTS** (already did, keep this going) -- ai/enabled defaults to false, so a fresh install makes no model calls at all. (already did, keep this going) -- ai/endpoint defaults to http://localhost:11434/api/chat, so the first thing it reaches for is a local model. (already did, keep this going) -- ai/provider defaults to ollama. (already did, keep this going) -- The plugin manifest declares network.access rather than assuming it. (already did, keep this going) -- **TESTING** + +- [ ] Create Rust<---> typescript communication layer **MUST BE IMPLEMENTED** + +- [ ] use **Nanocoder as the backend AI layer** instead of current existing , roughly sketched AI layer + +- [ ] **exclude Nanocoder existing TUI** + +- [ ] UI must **stay C++/QT** + +- [ ] all backend must route through **RUST BACKEND LAYER, INCLUDE NANOCODER-AI PARTS** (already did, keep this going) + +- [ ] ai/enabled defaults to false, so a fresh install makes no model calls at all. (already did, keep this going) + +- [ ] ai/endpoint defaults to http://localhost:11434/api/chat, so the first thing it reaches for is a local model. (already did, keep this going) + +- [ ] ai/provider defaults to ollama. (already did, keep this going) + +- [ ] The plugin manifest declares network.access rather than assuming it. (already did, keep this going) + +- [ ] **TESTING** before v1 ## could-do(s) From f34a1286adee54bb73661e3e5ee94464480f9be9 Mon Sep 17 00:00:00 2001 From: jason1015-coder Date: Mon, 17 Aug 2026 10:39:11 +0100 Subject: [PATCH 2/4] Revise must-do list for v1 in scriptura.md Updated the must-do list for v1 requirements in scriptura.md, including changes to AI layer, backend routing, and defaults for AI settings. --- content/collective/whitepapers/scriptura.md | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) diff --git a/content/collective/whitepapers/scriptura.md b/content/collective/whitepapers/scriptura.md index bfeec93..28f2322 100644 --- a/content/collective/whitepapers/scriptura.md +++ b/content/collective/whitepapers/scriptura.md @@ -139,33 +139,21 @@ Questions 1 (naming) and 2 (default provider) were settled during the review win Must exist in v1 and after throughout: - +- [ ] update the apple specfic secrets for nanocoder for a worked MacOS app registery - [ ] **MUST update** docs of scriptura to match Nano-collective's brand guidelines - - [ ] find some contributors, at least **one more core maintainer** (**jason1015-coder alone not practical to do all work**) by a issue in the transferred repo to Nano-collective - - [ ] mainwindow.cpp:888 reads settings straight out of QSettings, **MUST CHANGE** to: - OS keychain - encrypt it (personal key) - - [ ] Create Rust<---> typescript communication layer **MUST BE IMPLEMENTED** - - [ ] use **Nanocoder as the backend AI layer** instead of current existing , roughly sketched AI layer - - [ ] **exclude Nanocoder existing TUI** - - [ ] UI must **stay C++/QT** - - [ ] all backend must route through **RUST BACKEND LAYER, INCLUDE NANOCODER-AI PARTS** (already did, keep this going) - - [ ] ai/enabled defaults to false, so a fresh install makes no model calls at all. (already did, keep this going) - - [ ] ai/endpoint defaults to http://localhost:11434/api/chat, so the first thing it reaches for is a local model. (already did, keep this going) - - [ ] ai/provider defaults to ollama. (already did, keep this going) - - [ ] The plugin manifest declares network.access rather than assuming it. (already did, keep this going) - - [ ] **TESTING** before v1 ## could-do(s) From 8e0702f4a618ac6458f6efe2fedec09a100bfccf Mon Sep 17 00:00:00 2001 From: jason1015-coder Date: Sat, 22 Aug 2026 12:50:43 +0100 Subject: [PATCH 3/4] Refine explanation of editor capabilities and context engine Clarified the explanation of the proprietary editor's capabilities and the significance of the context engine and agent loop. --- content/collective/whitepapers/scriptura.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/collective/whitepapers/scriptura.md b/content/collective/whitepapers/scriptura.md index 28f2322..d4a24f1 100644 --- a/content/collective/whitepapers/scriptura.md +++ b/content/collective/whitepapers/scriptura.md @@ -67,7 +67,7 @@ All outbound AI network calls are routed through the Rust backend, where the per ### Why this shape is the point -A proprietary editor could do steps 2-4 with a better model. What it cannot do is step 5 by default, or let the developer swap the model in steps 2-4 for one they own. The combination of "Cursor feel" plus "local by default" plus "any model behind a real contract" is the part no one has shipped openly. The context engine and the agent loop are the hard parts; the provider abstraction is what makes the whole thing mean something. +The combination of "Cursor feel" plus "local by default" plus "any model behind a real contract" is the part no one has shipped openly. The context engine and the agent loop are the hard parts; the provider abstraction is what makes the whole thing mean something. ### Context engine From 254b7e213860ac93894ee8e80f3cfd4a4ecbe328 Mon Sep 17 00:00:00 2001 From: jason1015-coder Date: Sat, 22 Aug 2026 12:56:41 +0100 Subject: [PATCH 4/4] Clarify Scriptura's capabilities and alternatives Updated the notations regarding the capabilities and limitations of Scriptura, clarifying its position in the market and alternatives considered. --- content/collective/whitepapers/scriptura.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/content/collective/whitepapers/scriptura.md b/content/collective/whitepapers/scriptura.md index d4a24f1..19dd25b 100644 --- a/content/collective/whitepapers/scriptura.md +++ b/content/collective/whitepapers/scriptura.md @@ -110,6 +110,7 @@ What v1 ships is "an open editor with the Cursor feel, a real provider contract, - **Not a Copilot replacement that phones home.** The default install makes no remote calls. Remote providers are opt-in configuration, never hidden behaviour. - **Not a model trainer or a model vendor.** Scriptura uses whichever providers the user points it at. The collective does not train or ship an editor-tuned model of its own in v1. - **Not a from-scratch editor.** It is built on the existing base (scriptura), maintainers should not . A clean-room reimplementation would forfeit that inheritance for no gain. +- **Not a competitor to VScode, Jetbrains IDEs, Cursor, windsurf, Kiro, or Devin etc. AT LEAST FOR v1** The editor will not be able to competing direct in the market in v1, although we aim for similar AI editing features - **Not a guaranteed-latency product on weak hardware.** Local-first means the feel depends on the local model. On a machine too small to run a completion model, the experience degrades; the project documents the floor rather than hiding it. - **Not a replacement for terminal agents.** Nanocoder in the terminal still wins for some workflows. Scriptura is the in-editor surface, not the only surface. - **Not a semantic retrieval product in v1.** The context engine uses lexical and symbol-aware search only. Embedding-based retrieval is a future idea, scoped out of v1 to keep the local-first promise honest and the implementation within reach. @@ -117,7 +118,7 @@ What v1 ships is "an open editor with the Cursor feel, a real provider contract, ## Alternatives considered -- **Fork Cursor directly.** Impossible: Cursor is closed source. Its value is in the proprietary layer we are precisely trying to replace. No fork path exists. +- **Fork Cursor, windsurf(devin desktop) directly.** Impossible: Cursor and windsurf are closed sourced. Its values are in the proprietary layers we are precisely trying to replace. No fork paths exist. - **Ship only as a VS Code extension, not a fork.** Already exists, but has less potential for expansion, integration, and customization( restricted by Microsoft's existing frame). - **Fork VS Code.** Possible, but more performance overhead (although classified as "lightweight" but not friendly toward normal users without extremely good hardware to run alongside with ollama or other local LLM providers) , which is not good for a machine already running a local LLM. - **Fork IntelliJ IDEA.** Even worst performance (heavy weight) and an even harder tech stack (Java-based), with an even more restricted architecture (forced java-based editor APIs) for expansion compared to VS Code, not favorable at all for local models.