docs: update clone instructions - #4
Open
BlakeEvans22 wants to merge 1 commit into
Open
Conversation
Open
MindXpansion
added a commit
to MindXpansion/gemma-chat
that referenced
this pull request
May 26, 2026
Adds research/01-mlx-vlm-and-gemma-4.md (440 lines), research/02-image- generation-strategy.md (530 lines), research/03-sota-electron-and-skills- architecture.md (789 lines), research/04-aios-integration-and-self- improvement.md (700 lines). §10 synthesizes the four investigations into a research-grounded revised roadmap that supersedes §9's first-principles thinking. Locks 8 cross-cutting architectural decisions: 1. Tool shape: Vercel AI SDK v5 (inputSchema/outputSchema), no `ai` dep 2. IPC validation: hand-rolled Zod defineHandler wrapper (NOT electron-trpc) 3. Skills format: Anthropic SKILL.md verbatim; reconcile with Bear's v3.1 when the standard docs are readable 4. Action wire format: stays as <action> XML — preserve the streaming-safe parser and partial-disk-write feature 5. MCP integration: Phase 2 not Phase 1 (SKILL.md first, MCP after approval surface lands) 6. Self-improvement: 3-tier append-only (per-conv → app-level → Bear-mediated master library), 3-data-point threshold per RISE 7. AIOS bridge: SESSION_HANDOFF_gemma_*.md files to claude-tracks/ Knowledge_Base/ — the canonical artifact 8. Temporal grounding: full system message (date/time/tz/workspace/ operator/runtime) on every conversation create Surfaces three conflicts requiring Bear's call: - write-boundaries (Bear chose "write freely"; Research ammaarreshi#4 recommends restricted v0.2 — both defensible, Bear's call stands but downshift recommended) - two governance docs (CLAUDE Standards, Skill v3.1) were Read-denied during research; Anthropic-format-first lets us ship without blocking - skills directory allowlist initial set Revised Phase 1 (6 items, ~1-2 days): Patch 5 multimodal, Patch 6 lying spinner, Patch 7 chat timeout, Patch 8 comment cleanup, audio gating + voice-via-MLX, initial AIOS integration. Revised Phase 2 (6 items, ~1-2 weeks): image gen tool (fal.ai), IPC Zod validation, Skills loader, run_bash hardening with typed CLIs + sandbox-exec, FS approval UX, session handoffs + pattern accumulation. Phase 3 (deferred): Electron Fuses, ASAR integrity, sandbox: true, notarization, hardened runtime, MCP client. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
MindXpansion
added a commit
to MindXpansion/gemma-chat
that referenced
this pull request
May 26, 2026
standard and CLAUDE.md Standard v1.0 Bear granted read access to the two governance docs. Updates §10: - §10.2.3: Replaces Research ammaarreshi#4's inferred frontmatter shape with the actual v3.1 spec. Confirms no `runtime:` or `model_min:` fields exist (Research ammaarreshi#4 proposed them speculatively) — Phase 2.3 uses the allowlist approach unambiguously. Notes that `allowed-tools` needs a Gemma-Chat-side mapping (Claude Code resolves to Desktop Commander; we resolve to our workspace-bounded equivalents). Clarifies Gemma Chat as a Skill RUNTIME only reads SKILL.md + frontmatter — the full v3.1 artifact (autoresearch loop, evaluation-criteria, references/, patterns/{program,evaluate,train,prepare}.* etc.) is build-time + Claude-Code-side. - §10.3.2: Marks resolved. Surfaces additional implications from the full read: MX1 (never repeat what LLMs know) applies to our system prompts; MX9 OUROBOROS co-management; MX4 Core Triad is Bear+Claude- only but temporal grounding is the piece Gemma Chat does adopt; v3.1 §10 Three-System Persistence requirement validates hybrid write-boundaries + Phase 2.6/2.7; v3.1 §6 Anti-Bloat applies to any SKILL.md we ever ship back. Skills loader (Phase 2.3) is now unblocked. neo4j-kg-architect sub-agent still researching Phase 2.7 RAG schema in background. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
ammaarreshi/gemma-chatcdcommand togemma-chatWhy
The README in this repository currently points new contributors at
ammaarreshi/gemma-chat-public, which can lead them to clone the public mirror instead of this canonical repository.