Capture Learnings
- 5 installs
- 4.4k repo stars
- Updated August 2, 2026
- builderio/agent-native
capture-learnings is a Claude skill that captures and applies accumulated knowledge in a learnings.md file across conversations.
About
This skill defines how the agent captures and applies accumulated knowledge in a learnings.md file. A developer relies on it so the agent records user preferences, corrections, and technical learnings, reads them before significant work, and updates them when it learns something worth remembering. It specifies what to capture, a category-grouped format, and when a learning should graduate into a skill.
- Captures user preferences and technical learnings in a learnings.md memory file
- Read before significant work and updated on feedback or corrections
- Defines what to capture, a grouped format, and a graduation path to skills
Capture Learnings by the numbers
- 5 all-time installs (skills.sh)
- Ranked #13,035 of 16,556 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 3, 2026 (Skillselion catalog sync)
capture-learnings capabilities & compatibility
- Capabilities
- memory · create skill
- Use cases
- memory
- Pricing
- Free
What capture-learnings says it does
Capture and apply accumulated knowledge in learnings.md.
**Read `learnings.md` before starting significant work.** Update it when you learn something worth remembering.
**learnings.md is gitignored** — safe for personal info, preferences, contacts
npx skills add https://github.com/builderio/agent-native --skill capture-learningsAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 5 |
|---|---|
| repo stars | ★ 4.4k |
| Last updated | August 2, 2026 |
| Repository | builderio/agent-native ↗ |
What it does
Maintain a learnings.md memory file so the agent remembers preferences and technical learnings across sessions.
Who is it for?
Recording user preferences, corrections, and technical gotchas for future sessions.
Skip if: Documenting things obvious from the code or standard framework behavior.
When should I use this skill?
The user gives feedback, shares preferences, corrects a mistake, or something is worth remembering.
What you get
A concise, category-grouped learnings.md the agent reads before work.
- updated learnings.md
Files
Capture Learnings
This is background knowledge, not a slash command. Read `learnings.md` before starting significant work. Update it when you learn something worth remembering.
When to Capture
User Preferences & Memory
- Tone and style — "I prefer casual tone", "don't use emojis", "keep replies short"
- Personal context — contacts, relationships, habits ("my wife's email is...", "I'm in PST timezone")
- Workflow preferences — "always CC my assistant", "I like to review before sending"
- Corrections — user says "no, do it this way instead" — capture the right way
Technical Learnings
- Surprising behavior — something didn't work as expected and you figured out why
- Repeated friction — you hit the same issue twice; write it down so there's no third time
- Architectural decisions — why something is done a certain way (the "why" isn't in the code)
- API/library quirks — undocumented behavior, version-specific gotchas
Don't Capture
- Things obvious from reading the code
- Standard language/framework behavior
- Temporary debugging notes
- Anything already in AGENTS.md or skills
Format
Add entries to learnings.md at the project root. Group by category:
## Preferences
- Prefers casual, direct tone — no corporate speak
- Always BCC assistant@company.com on client emails
- Wife's email: jane@example.com — reference as "Jane"
## Technical
- Apollo API returns null for personal Gmail addresses, only works for work emails
- HubSpot lifecycle stages are lowercase in the API but title case in the UI
## Patterns
- When drafting replies, match the sender's formality level
- For investor emails, keep it under 3 paragraphsKey Rules
1. Read first, write second — always check learnings.md before starting work 2. Capture immediately — don't wait until the end of the conversation 3. Keep it concise — one line per learning, grouped by category 4. Don't duplicate — if a learning exists, refine it rather than adding another 5. learnings.md is gitignored — safe for personal info, preferences, contacts
Graduation
When a learning is referenced repeatedly, it may belong in AGENTS.md or a skill:
- Updating
learnings.mdis a Tier 1 modification (data — auto-apply) - Updating a SKILL.md based on learnings is Tier 2 (source — verify after)
Related Skills
- self-modifying-code — learnings.md updates are Tier 1; skill updates are Tier 2
- create-skill — when a learning graduates, create a skill from it
Related skills
Setup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196k
Lark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402k
CavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7k
Lark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16k
Running Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72
Codex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10FAQ
When should the agent read the file?
Always check learnings.md before starting significant work.
Is learnings.md committed?
No, learnings.md is gitignored, so it is safe for personal info and preferences.