spideynolove/sequential-thinking-mcp-v2
Instructions file
Instructions for spideynolove/sequential-thinking-mcp-v2, covering claude.md, current state, development notes, target architecture boundaries and memory layers.
Command Claude Code
Compare code edits against stored rules to detect violations.
Command Claude Code
Daily health check for the memory system.
Command Claude Code
Review tracked rule violations before starting work.
Command Claude Code
Create a structured diary entry from the current session transcript.
Command Claude Code
Multi-dimensional analysis of a diary entry or session event to extract high-quality memories.
Command Claude Code
Weekly comprehensive audit of the memory system.
Command Claude Code
Analyze diary entries to identify patterns and propose CLAUDE.md updates.
Hook Claude Code
Runs before the context is compacted, executing pre_compact.py via python. From spideynolove/memory.
Hook Claude Code
Runs when you submit a prompt, before the agent sees it, executing user_prompt_submit.py and context_router.py via python (2 commands). From spideynolove/memory.
Settings file Claude Code
Agent settings declaring 2 hook events (PreCompact, UserPromptSubmit).
Agent
Analyzes a codebase using repomix + sequential-thinking + knowledge-graph. Returns structured tables covering architecture, runtime behaviors, and failure modes. Stores findings in .aim/ for future sessions. Use when onboarding to a new codebase or before planning a feature.
Agent
Executes MCP tool calls and CLI tasks while keeping tool schemas out of the main context window. Tries available AI CLIs in priority order before falling back to direct mcporter calls. Use when any agent needs to run an MCP tool without polluting its own context.
Agent
Runs the full coding workflow end-to-end for a given goal: clarify → analyze codebase → plan → implement → review → test. Spawns codebase-analyst and mcp-manager as subagents. Use when given a feature request or bug fix that requires understanding the codebase before touching code.
Agent
Generic task executor that delegates work to a specified AI CLI tool based on role assignment. Receives structured task with role, tool preference, and context from predecessor tasks. Use when the orchestrator needs to delegate a task to an external tool.
Skill Claude CodeCodex
Use when starting work on a codebase you want to remember across sessions, extracting architectural facts, or querying persistent project knowledge instead of re-reading files. Centers the knowledge-graph MCP and project-local .aim storage.
Skill Claude CodeCodex
Use code-review-graph through CLI or mcporter for codebase maps, impact analysis, semantic search, graph status, and review context without keeping the MCP server natively registered in Codex.
Skill Claude CodeCodex
Use when the user is learning a new concept, wants to build skill through interaction, or asks for explanations alongside solutions. Guides Codex toward high skill-formation interaction modes instead of pure delegation.
Skill Claude CodeCodex
/home/hung/.claude/skills/next-best-practices/SKILL.md.
Skill Claude CodeCodex
/home/hung/.claude/skills/nuxt/SKILL.md.
Skill Claude CodeCodex
/home/hung/.claude/skills/quality-tests/SKILL.md.
Skill Claude CodeCodex
Orchestrates the RLM repo workflow end-to-end with phase gates, locked artifacts, addenda, traceability, and automatic bootstrap/upsert of AGENTS/PLANS scaffolding. Trigger phrases: "Implement requirement ", "Run RLM Phase ", "resume requirement", "lock Phase ", "verify locks".
Agent
Use this agent when a major project step has been completed and needs to be reviewed against the original plan and coding standards. Examples: After sub-phase implementation, after major feature completion, before merging to verify work meets requirements.
Agent
Use this agent when implementing a single sub-phase from an RLM plan. The agent should receive the full sub-phase text, current codebase context, and implement the feature with TDD discipline.