Joseph Mearman Mearman

Claude Flow treats memory as the backbone and MCP tools as the hands. You get concurrent agents that coordinate cleanly, keep context tight, and ship durable artifacts without dragging long text through prompts. It feels like an ops layer for intelligence.

The stack is simple. Claude Code as the client. Claude Flow as the MCP server. SQLite memory at .swarm/memory.db for state, events, patterns, workflow checkpoints, and consensus. Artifacts hold the big payloads. Manifests in memory link everything with ids, tags, and checksums.

Coordination is explicit. Agents write hints to a shared blackboard, gate risky steps behind consensus, and record every transition as an event. Hooks inject minimal context before tools run and persist verified outcomes after. Small bundles in, durable facts out.

Planning keeps runs stable. Use GOAP to sequence actions with clear preconditions. Use OODA to shorten loops.

Observe metrics, orient with patterns, decide through votes, act with orchestration. Topology adapts from hi

Mode important instructions:

IMPORTANT NOTE: Adherence to the rules listed in this Mode important instructions block, (e.g: MEMORY BANK COLLABORATION), adherence to the rules takes precedence and should not be forgotten.

MEMORY BANK COLLABORATION:

NOTE: While in this preload process, communication with the user should be short, concise and to the point, e.g: Memory setup found, Do you want to preload the memory bank?, etc.

Check: Try to do a quick check to see if other modes instructions have any kind of memory bank setups, prioritize the setup of default mode if found.
Additional check: If no or multiple setups pattern was found during step 1, look for a memory bank setup at the root repository level, refer to #known-setups for more details, proceed with the setup that is most relevant.

Which GGUF is right for me? (Opinionated)

Good question! I am collecting human data on how quantization affects outputs. See here for more information: ggml-org/llama.cpp#5962

In the meantime, use the largest that fully fits in your GPU. If you can comfortably fit Q4_K_S, try using a model with more parameters.

llama.cpp feature matrix

See the wiki upstream: https://github.com/ggerganov/llama.cpp/wiki/Feature-matrix

Common special files found in the root directory of a repository

Description for and list of popular special files like README/CHANGELOG/LICENSE and others.

README-like

ReadMe `README.md` `README`

The ReadMe is usually the first document people will see of your project. Depending on your project it should give a short introduction and usage/build examples. It should only contain the information you expect users to read. It is usually possible to link to other documentation files using the markdown syntax which gets rendered as html by popular repository hosting platforms.

How to install game-porting-toolkit (aka proton for macOS)

You also might wanna just use Whisky which does this automatically

This guide works on macOS 13.4+ using Command Line Tools for XCode 15 Beta!

What is this?

In the recent WWDC, Apple announced and released the "game porting toolkit", which upon further inspection this is just a modified version of CrossOver's fork of wine which is a "compatibility layer" that allows you to run Windows applications on macOS and Linux.

This worked on 14/May/23. The instructions will probably require updating in the future.

llama is a text prediction model similar to GPT-2, and the version of GPT-3 that has not been fine tuned yet. It is also possible to run fine tuned versions (like alpaca or vicuna with this. I think. Those versions are more focused on answering questions)

Note: I have been told that this does not support multiple GPUs. It can only use a single GPU.

It is possible to run LLama 13B with a 6GB graphics card now! (e.g. a RTX 2060). Thanks to the amazing work involved in llama.cpp. The latest change is CUDA/cuBLAS which allows you pick an arbitrary number of the transformer layers to be run on the GPU. This is perfect for low VRAM.

Clone llama.cpp from git, I am on commit 08737ef720f0510c7ec2aa84d7f70c691073c35d.

Chat GPT "DAN" (and other "Jailbreaks")

	{
	"customModes": [
	{
	"slug": "sparc",
	"name": "⚡️ SPARC Orchestrator",
	"roleDefinition": "You are SPARC, the orchestrator of complex workflows. You break down large objectives into delegated subtasks aligned to the SPARC methodology. You ensure secure, modular, testable, and maintainable delivery using the appropriate specialist modes.",
	"customInstructions": "Follow SPARC:\n\n1. Specification: Clarify objectives and scope. Never allow hard-coded env vars.\n2. Pseudocode: Request high-level logic with TDD anchors.\n3. Architecture: Ensure extensible system diagrams and service boundaries.\n4. Refinement: Use TDD, debugging, security, and optimization flows.\n5. Completion: Integrate, document, and monitor for continuous improvement.\n\nUse `new_task` to assign:\n- spec-pseudocode\n- architect\n- code\n- tdd\n- debug\n- security-review\n- docs-writer\n- integration\n- post-deployment-monitoring-mode\n- refinement-optimization-mode\n\nValidate:\n✅ Files < 500 lines\n✅ No hard-coded

	customModes:
	- slug: security-auditor
	name: 🛡️ Security Auditor
	roleDefinition: Act as an expert security researcher conducting a thorough
	security audit of my codebase. Your primary focus should be on identifying
	and addressing high-priority security vulnerabilities that could lead to
	system compromise, data breaches, or unauthorized access.
	customInstructions: >-
	Follow this structured approach: