improve-skills

Updates existing skills with debugging patterns and workflow learnings captured from the current session.

1|Updated May 13, 2026
One-click install
npx skills add https://github.com/dlt-hub/dlthub-start --skill improve-skills-dlt-hub
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: improve-skills
Source: https://github.com/dlt-hub/dlthub-start/tree/main/src/create_dlthub_workspace/scaffolds/minimal_workspace/_agents/cursor/.cursor/skills/improve-skills
Command: npx skills add https://github.com/dlt-hub/dlthub-start --skill improve-skills-dlt-hub

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Valuable lessons learned during a working session — debugging fixes, data quirks, useful doc links — are usually lost when the session ends. This Skill captures those learnings and folds them back into your team's skill files so institutional knowledge accumulates instead of evaporating. ## Core Features & Use Cases - Session Learning Extraction: Scans the conversation for new error patterns, schema surprises, query techniques, and configuration quirks worth preserving. - Lean, Targeted Updates: Reads each relevant SKILL.md first, avoids duplication, and appends only minimal additions such as a bullet point or a short code snippet. - Approval-Gated Edits: Presents a summary of proposed changes per skill and waits for user approval before modifying any files. - Use Case: After a long session debugging a dlt pipeline pagination issue, run this Skill to add the root cause, the diagnostic MCP tool used, and an essential docs link to the relevant skill so the next session starts smarter. ## Quick Start Ask the agent to review this session and update the relevant skills with anything new we learned.

Frequently Asked Questions about improve-skills

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I update AI agent skills after a working session?▼

Run the improve-skills workflow at the end of the session. It scans the conversation for new learnings, maps them to the relevant SKILL.md files, proposes a summary of changes, and applies them only after your approval.

What kinds of learnings get captured in skill updates?▼

The skill captures new error types and their diagnostic commands, data type and schema surprises, useful query patterns, source configuration quirks like auth or pagination behavior, and essential documentation links discovered during the session.

Does improve-skills rewrite or restructure existing skill files?▼

No. It reads each target SKILL.md first, avoids duplicating existing content, and only appends minimal additions to the most relevant section. It never reorganizes or rewrites existing skill content.

Can I review changes before skills are modified?▼

Yes. The skill presents a per-skill summary of proposed updates and waits for explicit user approval before editing any files. You can adjust, skip, or extend the proposed changes at that point.

When should I not use session-based skill updates?▼

Avoid it for hypothetical scenarios or generic best practices not encountered in the session. The skill is designed for real, observed problems and explicitly requested captures, so speculative additions dilute the skill files.