Research Resources

Resources that contain research that can be used to build plugins.

List

[]YOLO Mode (You Only Look Once) automates your entire Phases workflow - Claude have --dangerously-skip-permissions flag to skip permissions check, so it can be used to run YOLO Mode without permissions check. []Agent0 - Unleashing Self-Evolving Agents from Zero Data via Tool-Integrated Reasoning

• https://github.com/aiming-lab/Agent0 []Solving a Million-Step LLM Task with Zero Errors - using cat file | claude -p "query" --output-format will run Query via SDK, then exit with json output.
• Adding --max-turns 3 will limit amount of turns to 3.
• --json-schema '{"type":"object","properties":{...}}' "query" will validate the output against a JSON schema.
• --model flag to specify the model to use.
• --permission-mode plan will run agent in specified permissions mode https://code.claude.com/docs/en/iam#permission-modes

claude --agents '{
  "code-reviewer": {
    "description": "Expert code reviewer. Use proactively after code changes.",
    "prompt": "You are a senior code reviewer. Focus on code quality, security, and best practices.",
    "tools": ["Read", "Grep", "Glob", "Bash"],
    "model": "sonnet"
  },
  "debugger": {
    "description": "Debugging specialist for errors and test failures.",
    "prompt": "You are an expert debugger. Analyze errors, identify root causes, and provide fixes."
  }
}'

[]Agent OS - Agent OS is a spec-driven development system that gives AI agents the structured context they need to write production-quality code. [x]codemap - map codebase structure []Effective harnesses for long-running agents []mgrep [x]arxhiv MCP []Docker MCP Toolkit [][Arc42 specification template] - Research Arc42 and adapt it for use in Spec Driven Development. []Opus soul document []YAGNI - Yagni originally is an acronym that stands for “You Aren't Gonna Need It”. It is a mantra from ExtremeProgramming. []Extreme Programming []Beads task traker cli - maybe better to create new cli with simplified architecture, that useses only TASKS.md file. []Building the 14 Key Pillars of Agentic AI [] Three of Thought and etc - Expand papers that used in reflect plugin as separate comamnds/skills/hooks []Building Reliable RAG Pipelines Is Still Hard In 2025 [] LSP server integration with Claude Code []Conductor: Google spec driven development kit [] Task tracking: https://github.com/hmans/beans https://github.com/rrnewton/minibeads https://github.com/steveyegge/beads [x]Agent Skills for Context Engineering []Agent search MCP []First Principles Framework [] Think about way to make reflection work as continues-learning agent. It can trigger on words like "You absolutily right", analyse it and save to CLAUDE.md file correction that user provided. []Hookify - advanced hook configuration, that using python skills. []Ralph - continus iteration plugin and orcestrator verision https://github.com/mikeyobrien/ralph-orchestrator []Security Reminder - hook that reminds about security best practices. [] Add git workspaces usage for competitive model writing [] Research how git notes can be used during code writing and review [] Research how to add RAG style pipline with vector search to prepent relevant code to context window before code writing [] Check "Prompting Science" series. https://arxiv.org/abs/2503.04818, https://arxiv.org/abs/2512.05858, https://chatpaper.com/paper/172346, https://arxiv.org/abs/2508.00614, https://www.researchgate.net/publication/392530384_Prompting_Science_Report_2_The_Decreasing_Value_of_Chain_of_Thought_in_Prompting [] https://arxiv.org/html/2602.16666v1 - Towards a Science of AI Agent Reliability [] https://arxiv.org/html/2601.06112v1 - ReliabilityBench: Evaluating LLM Agent Reliability Under Production-Like Stress Conditions