Distill Memory
Persist decisions, procedures, and lessons from agent sessions into searchable memories so solo builders can resume work without re-explaining context.
Overview
Distill Memory is a journey-wide agent skill that captures decisions and insights into searchable knowledge-base memories—usable whenever a solo builder needs to preserve context before a session ends.
Install
npx skills add https://github.com/nowledge-co/community --skill distill-memoryWhat is this skill?
- Proactively saves decisions, preferences, plans, procedures, and learnings—no need to ask the user first
- Distinguishes add vs update to avoid duplicate memories on the same topic
- Structured saves with unit types: decision, procedure, learning, preference, event
- Importance scoring (0.8–1.0 for major insights) plus labels for retrieval
- Explicit end-of-task check to distill one durable memory from substantial work
- 5 structured unit types: decision, procedure, learning, preference, event
- Importance scale 0.8–1.0 for major insights
Adoption & trust: 615 installs on skills.sh; 91 GitHub stars; 3/3 security scanners passed (skills.sh audits).
What problem does it solve?
You finish a long agent session full of decisions and debugging lessons, then the next chat starts cold with no durable record of what you learned.
Who is it for?
Solo builders running multi-session agent work who use or plan a memory tool (`nmem`) and want automatic distillation of durable decisions and procedures.
Skip if: One-off Q&A, throwaway tasks with no reusable lesson, or teams that already enforce a single source of truth in a wiki without wanting agent-side memory capture.
When should I use this skill?
Conversation produces a decision, preference, plan, procedure, learning, or important context—or at the end of a substantial task.
What do I get? / Deliverables
Important context lands as labeled, typed memories you can search and update, so later sessions resume with shared understanding instead of repeated explanation.
- New or updated searchable memory entries with titles, labels, and unit types
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
Record why you narrowed MVP scope so a later agent session does not reopen settled tradeoffs.
Save the PostgreSQL vs SQLite decision with ACID rationale as a decision-unit memory.
Distill a repeatable PR checklist procedure after a painful review cycle.
Capture incident root cause and fix steps as a learning memory for on-call reuse.
Store durable support macros or escalation preferences support agents should honor.
How it compares
Use as procedural capture discipline alongside ad-hoc chat history—not as a replacement for formal docs or issue trackers.
Common Questions / FAQ
Who is distill-memory for?
Solo and indie builders using coding agents who want searchable memories for decisions, workflows, and lessons instead of losing context between sessions.
When should I use distill-memory?
During build when you lock an architecture choice, after ship/operate debugging when you have a root cause, during validate when pricing or scope decisions need rationale, or anytime a substantial task ends and one durable memory would help the next session.
Is distill-memory safe to install?
Review what the agent writes into your knowledge base and any CLI access implied by `nmem`; check the Security Audits panel on this Prism page before enabling in production repos with secrets.
SKILL.md
READMESKILL.md - Distill Memory
# Distill Memory Save proactively when the conversation produces a decision, preference, plan, procedure, learning, or important context. Do not wait to be asked. ## When to Save Good candidates include: - decisions with rationale ("we chose PostgreSQL because ACID is required") - repeatable procedures or workflows - lessons from debugging, incidents, or root cause analysis - durable preferences or constraints - plans that future sessions will need to resume cleanly - important context that would be lost when the session ends Skip routine fixes with no generalizable lesson, work in progress that will change, simple Q&A answerable from documentation, and generic information already widely known. ## Add vs Update - Use `nmem --json m add` when the insight is genuinely new. - If an existing memory already captures the same decision, workflow, or preference and the new information refines it, use `nmem m update <id> ...` instead of creating a duplicate. - At the end of a substantial task, explicitly check whether one durable memory should be added or updated. Prefer atomic, standalone memories with strong titles and clear meaning. Focus on what was learned or decided, not routine chatter. Use structured saves when possible: `--unit-type` (decision, procedure, learning, preference, event), `-l` labels, `-i` importance (0.8–1.0 major decisions, 0.5–0.7 useful patterns, 0.3–0.4 minor notes). ## Native Connector These skills work in any agent via CLI. For auto-recall, auto-capture, and graph tools, check if your agent has a native Nowledge Mem connector — run the `check-integration` skill or see https://mem.nowledge.co/docs/integrations