Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Claude Code starts every conversation from scratch.
It doesn't remember your project conventions, doesn't know why you chose approach A last time, and won't learn from past failures. You'll fall into the same traps over and over again.
Pensieve gives Claude Code structured memory. Every review, every commit, every complex task automatically accumulates project knowledge. The longer you use it, the better it understands your project.
| Without | With |
|---|---|
| Re-explain project conventions every time | Conventions stored as maxims, loaded automatically |
| Complex tasks spiral out of control halfway through | Loop auto-decomposes, isolates execution, verifies each step |
| Code review standards depend on your mood that day | Review standards hardened into executable pipelines |
| Same mistake from last week happens again this week | Lessons learned auto-distilled, skipped next time |
| Forget why a technical decision was made three months later | Decision records include context and mitigation checklists |
Day 1 — Install → Initialize → Auto-scan project hotspot modules → Output code-taste baseline report
Week 1 — Use loop to tackle complex dev tasks. Claude decomposes work according to your maxims, sub-agents execute in isolation, and lessons are auto-distilled at wrap-up.
Month 1 — Your project has accumulated its own conventions, technical decision records, review workflows, and reference knowledge. Every commit and review silently enriches this knowledge base.
After that — Claude understands your project better and better. When new team members join, Pensieve serves as a living project handbook.
# 1. Add the marketplace source
claude plugin marketplace add kingkongshot/Pensieve#main
# 2. Install
claude plugin install pensieve@kingkongshot-marketplace --scope user
# 3. Restart Claude Code, then say:
Initialize pensieve for me
That's it. Then say "use loop to finish this task" to kick off your first task.
Installation Guide · Update Guide · Uninstall
Ready to use after install, no extra configuration needed. Just describe what you want in plain language.
init — Initialize a ProjectScans your git history, identifies hotspot modules, and runs a code-taste baseline analysis. Creates the project-level knowledge directory (maxims / decisions / knowledge / pipelines) and seeds default review and commit pipelines. Analyze-only, no writes — you decide which findings are worth keeping.
Must run doctor once after completion for structure and format verification.
"Initialize pensieve for me"
loop — Decompose and Execute Complex TasksBreaks a large requirement into sub-tasks, confirms scope before starting. The main window orchestrates while sub-agents execute each task in isolation, keeping contexts clean. At wrap-up, automatically asks whether to distill lessons learned. Small tasks skip loop and run directly.
"Use loop to finish this task"
self-improve — Distill Lessons LearnedExtracts insights from conversations, diffs, and loop executions. Classifies them as maxim (hard rule), decision (technical decision), knowledge (reference fact), or pipeline (executable workflow), writes to the corresponding location, and updates the knowledge graph. Also auto-triggers on commit.
"Distill lessons from this session"
doctor — Health CheckRead-only scan of all user data: frontmatter format, semantic link integrity, directory structure compliance. Outputs a fixed-format PASS / PASS_WITH_WARNINGS / FAIL report with a three-step action plan. Does not modify user data files by default; only auto-maintains the SKILL.md and Claude auto memory (~/.claude/projects/<project>/memory/MEMORY.md) Pensieve guidance blocks.
"Check if there are any data issues"
upgrade — Version UpgradeOnly performs version-level actions: version comparison, pull latest, plugin key alignment and old plugin name cleanup. Does not perform directory migration or content repair. After upgrade completes, guides you to manually run doctor for verification.
"Upgrade pensieve"
migrate — Structure Migration and Residue Cleanupnpx claudepluginhub mamajiaa/pensieve-claude-plugin --plugin pensieveCorca Workflow Framework — consolidated hooks and skill orchestration for structured development sessions
OpenLTM — Long-Term Memory for AI coding agents: semantic search, context injection, session learning
Persistent memory system for AI coding sessions — cross-tool memory sharing with 6-dimensional hybrid search
Hand off context to any AI coding agent. Creates comprehensive handoff documents that capture the current task, progress, modified files, and all relevant context needed to seamlessly continue work in any AI assistant (Claude Code, Cursor, Copilot, Aider, etc.).
Plugin for effective agentic development
Plan and autonomously build a software task end-to-end. Recons the codebase, applies preloaded memory, decomposes into the right number of phases, gets one confirmation, then prepares a single ready-to-paste /goal command — one paste between you and done — that drives execution to completion with built-in retry, fix-spec recovery, and per-phase memory writeback. Works on Claude Code and Codex.