iamtouchskyer/memex/memex-retro
Claude Code and agent skill by iamtouchskyer/memex
- community
Checking that the SKILL.md file is still in the repository.
Save insights from completed tasks to Zettelkasten memory.
Install
npx vskill@latest install iamtouchskyer/memex/memex-retroSKILL.md
Memory Retro
You have access to a Zettelkasten memory system via the memex CLI. After completing this task, reflect on what you learned and save valuable insights.
Tools Available
Three equivalent interfaces — use whichever your environment supports:
| CLI (memex in PATH) | Plugin CLI fallback (Claude Code) | MCP tool (VSCode / Cursor) | |
|---|---|---|---|
memex search <q> | node ~/.claude/plugins/cache/cc-plugins/memex/*/dist/cli.js search <q> | memex_search with query arg | |
memex read <slug> | node ~/.claude/plugins/cache/cc-plugins/memex/*/dist/cli.js read <slug> | memex_read with slug arg | |
memex write <slug> | node ~/.claude/plugins/cache/cc-plugins/memex/*/dist/cli.js write <slug> | memex_write with slug arg and body |
Resolution order: Try memex in PATH first. If not found, define a shell function and use it:
If both CLI approaches fail, use MCP tools (memex_search, memex_read, memex_write).
The rest of this skill uses memex CLI syntax for brevity.
Process
- Ask yourself: what did I learn from this task that would be useful in the future?
- If nothing worth remembering, skip — not every task produces insights
- For each insight, draft an atomic card (one insight per card)
- Fact Hygiene Check — before writing, scan the draft for implicit context that a stranger (or future AI) couldn't decode. Ask three questions:
- WHO: Every project/product/team mentioned — is it the user's own work or external? Would a reader with zero context know?
- WHAT-WHEN: Every number (days, tokens, cost) — is it bound to a specific project name and time period?
- RELATIONSHIP: Words like "对标/参照/基于/借鉴/reference" — spell out the actual relationship (authored, benchmarked against, forked from, inspired by, etc.)
- If any answer is "a stranger couldn't tell", fix the draft before writing. One sentence of context prevents hallucinated narratives downstream.
- Before writing,
memex searchfor related existing cards - Dedup check: If an existing card already covers this insight,
memex readit, then update it by appending new information (usememex writewith the full updated content) - If it's genuinely new, write a new card with
[[links]]to related cards in the prose
Card Format
Note: You do NOT need to include modified — the CLI auto-sets it on write.
Rules
- Atomic: One insight per card. If you have 3 insights, write 3 cards.
- Own words: Don't copy-paste. Distill and rephrase — this is the Feynman method.
- Link in context:
[[links]]must be embedded in sentences that explain the relationship. - Good: "This contradicts what we found in [[jwt-migration]] — stateless tokens can't be revoked."
- Bad: "Related: [[jwt-migration]]"
- Slug: English kebab-case, descriptive. e.g.,
jwt-revocation-blacklist-pattern - Don't over-record: Only save insights that would change how you approach a similar task in the future.
- Preserve frontmatter on update: When updating an existing card, preserve its original frontmatter fields (title, created). Only append to the body.
sourceis auto-managed by the MCP server. - No raw secrets: Never write actual secrets, credentials, tokens, or exact secret file contents into memory. Use redacted examples and abstract descriptions instead.
External Source Cards (Flomo, etc.)
When curating cards from external sources like flomo:
- Use
source: flomo(notsource: retro). This is critical for anti-loopback guards. - Digest, don't copy. Read the raw memo, extract the genuine insight, rewrite as atomic Zettelkasten card.
- Merge multiple related memos into one card when they're about the same topic.
- Skip low-quality fragments — the bar for external imports is higher than session retro.
- These cards follow the same quality rules above (atomic, own words, linked).
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/memex-retro/SKILL.md
Popularity
- Repository stars
- 129
- Repository forks
- 38
- 7-day trend score
- 24
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/iamtouchskyer/memex/memex-retro)