knowledge-consolidation

Original：🇺🇸 English

Translated

1 scripts

Proactively summarize and consolidate knowledge from AI conversation sessions. Auto-triggers when: (1) Starting a new session after meaningful previous work, (2) Session contains significant learnings worth preserving. Captures debugging insights, architecture decisions, patterns, configs, and lessons learned into structured knowledge documents. Explicit triggers: 'summarize', 'consolidate', 'save knowledge', 'document this'.

3installs

Sourcelearnwy/skills

Added on2026-02-25

NPX Install

npx skill4agent add learnwy/skills knowledge-consolidation

SKILL.md Content

View Translation Comparison →

Knowledge Consolidation

Persist valuable knowledge from AI conversations into structured documents.

When to Invoke

Proactive (AI self-invoke):

Starting new session after substantial prior work
Accumulated debugging/architecture/pattern knowledge worth preserving
Resolved non-trivial issues with reusable insights

Skip when:

Simple Q&A with no novel insights
Trivial changes with nothing worth preserving

Workflow

Step 1: Detect AI IDE Type

Check project root for AI IDE indicator directory:

Environment Indicator	AI Type	Storage Path
`.trae/` dir or TraeAI/TraeCN env	`trae-cn`	`.trae/knowledges/`
`.claude/` dir or Claude Code env	`claude-code`	`.claude/knowledges/`
`.cursor/` dir	`cursor`	`.cursor/knowledges/`
`.windsurf/` dir	`windsurf`	`.windsurf/knowledges/`

Step 2: Identify Consolidation Candidates

Scan conversation for knowledge worth preserving:

Debug sessions with root cause analysis
Architecture decisions with rationale
Code patterns discovered or established
Configuration knowledge (build, env, tooling)
API integration insights
Workflow improvements
Lessons learned from issues

Step 3: Classify Knowledge Type

Select appropriate type based on content:

Type	When to Use
`debug`	Bug fixes, crash analysis, error resolution
`architecture`	System design, module structure, dependencies
`pattern`	Reusable code patterns, idioms, best practices
`config`	Build settings, environment, tooling setup
`api`	API design, integration, protocol details
`workflow`	Development processes, procedures
`lesson`	Post-mortems, retrospectives, project insights
`reference`	Technical specs, schema docs, standards

See knowledge-types.md for detailed guidance on each type.

Step 4: Generate Document Path

Run path generation script:

bash

{skill_root}/scripts/get-knowledge-path.sh \
  -r <project_root> -a <ai_type> -t <type> -n <filename>

Output format:

{project_root}/{ai_path}/knowledges/{YYYYMMDD}_{seq}_{type}_{filename}.md

Step 5: Write Knowledge Document

Create document using template:

markdown

# {Title}

> **Type:** {type}
> **Date:** {YYYY-MM-DD}
> **Context:** {Brief context}

## Summary

{2-3 sentence summary}

## Background

{Situation/problem/context}

## Details

{Technical content, code snippets, analysis}

## Key Takeaways

{Bullet points of actionable insights}

## Related

{Links to related files or knowledge entries}

Step 6: Confirm with User

📚 Knowledge Consolidated:
- File: {path}
- Type: {type}
- Topic: {title}

Summary: {brief summary}

Resources

Resource	Purpose
`scripts/get-knowledge-path.sh`	Generate timestamped file path
`references/knowledge-types.md`	Detailed type selection guide
`assets/knowledge.md.template`	Document template
`examples/debug-session.md`	Complete workflow example