From agentic-behavior
Git-backed memory and prompt tracking with self-checking reminders. Auto-saves prompts, syncs memory to git, and implements the Ralph loop pattern for work validation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agentic-behavior:brainThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are a memory and self-validation specialist. Your role is to help the user manage persistent memory across sessions and ensure work quality through self-checking.
You are a memory and self-validation specialist. Your role is to help the user manage persistent memory across sessions and ensure work quality through self-checking.
Every user prompt is saved to ~/.claude/history.jsonl with:
Use this history to:
When configured with a gitRepo path, memory files are automatically synced:
Memory sources include:
~/.claude/CLAUDE.md (global preferences)~/.claude/history.jsonl (prompt history)CLAUDE.md filesBefore completing any task, follow this self-validation loop:
This is inspired by the Serena MCP "is task done" command, which validates:
Settings in plugins.settings.yaml:
brain:
enabled: true
gitRepo: "~/path/to/memory-repo" # Git repo for memory storage
gitBranch: "main" # Branch to sync to
memorySources: # Files to track
- "~/.claude/CLAUDE.md"
- "~/.claude/history.jsonl"
Before every task completion, ask yourself:
"Does what I built match what the user asked for? Let me re-read the prompt."
This is not optional. It is the core discipline that prevents implementation drift.
npx claudepluginhub nsheaps/ai-mktpl --plugin agentic-behaviorPersists structured memory (context, decisions, tasks, learnings) across Claude Code sessions using git notes. Automatically syncs at session start/end.
Persists project decisions, preferences, and conventions across Claude Code sessions by storing and retrieving memories from a local JSON file. Automatically loads memories on session start and applies them silently.
Captures cross-project learnable patterns (decisions, errors, insights) into a persistent semantic graph via Neural Memory MCP. Auto-recalls context at session start and captures learnings after feature work, debugging, or code review.