gpd-record-insight

Record project-specific learnings and error patterns to a structured insights ledger.

Updated Mar 18, 2026
One-click install
npx skills add https://github.com/MichaelsEngineering/sentinel-multiscale-field-dynamics --skill gpd-record-insight-michaelsengineering
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gpd-record-insight
Source: https://github.com/MichaelsEngineering/sentinel-multiscale-field-dynamics/tree/main/.agents/skills/gpd-record-insight
Command: npx skills add https://github.com/MichaelsEngineering/sentinel-multiscale-field-dynamics --skill gpd-record-insight-michaelsengineering

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Research agents repeatedly rediscover the same mistakes and patterns across project phases. This Skill captures project-specific lessons, error patterns, and verification insights into a persistent .gpd/INSIGHTS.md ledger so future work avoids repeating known pitfalls. ## Core Features & Use Cases - Structured Insight Recording: Appends categorized table rows with date, phase, category, confidence level, description, and prevention guidance to .gpd/INSIGHTS.md. - Duplicate Detection: Searches the existing ledger before writing to prevent recording the same insight twice. - Automatic Section Routing: Maps insight categories (error-pattern, convention-pitfall, approximation-lesson, and others) to the correct ledger section and commits the change to git. - Use Case: After discovering that finite-size corrections scale as 1/L^2 rather than 1/L for a given geometry, record the finding as a verification lesson so later phases apply the correct scaling. ## Quick Start Ask the agent to record an insight describing the lesson learned, its category, and how to prevent it from recurring.

Frequently Asked Questions about gpd-record-insight

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

FAQPage Schema
How do I record a project insight or lesson learned?▼

Invoke the record-insight workflow with a description of the lesson. It creates `.gpd/INSIGHTS.md` if missing, checks for duplicates, appends a structured table row with date, phase, category, confidence, description, and prevention, then commits the change to git.

What categories of insights can be recorded?▼

Supported categories are error-pattern, debugging-pattern, convention-pitfall, approximation-lesson, computational-insight, and verification-lesson. Each category maps to a ledger section such as Debugging Patterns, Verification Lessons, Consistency Issues, or Execution Deviations.

Does the workflow prevent duplicate insights?▼

Yes. Before appending, the workflow greps the existing `.gpd/INSIGHTS.md` for keywords matching the new insight. If a match is found, it reports that the insight is already recorded and skips the write.

What does the confidence field mean for recorded insights?▼

Confidence is high, medium, or low, reflecting how confirmed the pattern is. High-confidence insights confirmed across two or more phases become candidates for promotion to the global pattern library at milestone completion.

Where are recorded insights stored?▼

Insights are stored in `.gpd/INSIGHTS.md` as Markdown tables organized by section. The file is created from a template on first use and each update is committed to git with a descriptive message.