From pensieve
Load this skill IMMEDIATELY when user expresses any intent. System capability (tools/knowledge/scripts) is shipped inside the plugin and updated only via plugin updates. User data must live in project-level `.claude/pensieve/` and is never overwritten by the plugin. When user wants to improve Pensieve system (plugin content), MUST use the Self‑Improve tool (`tools/self-improve/_self-improve.md`).
How this skill is triggered — by the user, by Claude, or both
Slash command
/pensieve:pensieveThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Route user intent to the right tool/pipeline.
decisions/README.mdknowledge/README.mdknowledge/taste-review/content.mdmaxims/README.mdpipelines/README.mdtools/loop/DESIGN.template.mdtools/loop/README.mdtools/loop/REQUIREMENTS.template.mdtools/loop/_loop.mdtools/loop/scripts/_lib.shtools/loop/scripts/end-loop.shtools/loop/scripts/find-task-list-id.shtools/loop/scripts/init-loop.shtools/loop/scripts/init-project-data.shtools/pipeline/_pipeline.mdtools/pipeline/scripts/list-pipelines.shtools/self-improve/_self-improve.mdtools/upgrade/_upgrade.mdtools/upgrade/templates/maxims.initial.mdtools/upgrade/templates/pipeline.review.mdRoute user intent to the right tool/pipeline.
skills/pensieve/
.claude/pensieve/
maxims/: your team principles (e.g. custom.md)decisions/: project decision recordsknowledge/: external references you addpipelines/: project pipelines (seeded on install)loop/: loop run outputs (one dir per loop)When to use:
Entry:
commands/loop.mdtools/loop/_loop.mdTriggers:
loop / "use loop"When to use:
Entry:
commands/selfimprove.mdtools/self-improve/_self-improve.mdTriggers:
When to use:
Entry:
commands/pipeline.mdtools/pipeline/_pipeline.mdTriggers:
When to use:
.claude/pensieve/Entry:
commands/upgrade.mdtools/upgrade/_upgrade.mdTriggers:
SessionStart injects the system capability path and project user‑data path into context as the single source of truth at runtime.
npx claudepluginhub michaeklishin/pensieve --plugin pensievePersists user decisions, constraints, and project memory across sessions into a local mdbase. Activates on architecture, tech stack, or UI/UX discussions.
Logs cross-project outcomes and recalls lessons to inform new sessions, avoiding past mistakes. Analyzes skill executions for better routing. Use /memento modes: log, global recall, health, route.
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.