Agent skill
knowledge-base
Use when recording ADRs, saving debug findings, persisting patterns, or referencing past decisions in docs/
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/knowledge-base-klara-copilot-epost-agent-kit
Metadata
Additional technical details for this skill
- keywords
-
knowledge adr decision pattern learning memory persist capture
- triggers
-
[ "ADR", "decision record", "learned pattern", "save finding" ] - platforms
-
[ "all" ] - agent affinity
-
[ "epost-architect", "epost-debugger", "epost-documenter", "epost-implementer" ]
SKILL.md
Knowledge Base Skill
Purpose
Shared project knowledge management system for persisting architectural decisions, learned patterns, debug findings, coding conventions, system architecture docs, and feature deep-dives across sessions.
When Active
- Recording architectural decisions
- Capturing implementation patterns
- Documenting debugging findings
- Establishing coding conventions
- Documenting system architecture
- Writing feature deep-dives
Directory Structure
docs/ # Project root (git-tracked)
├── index.json # Machine-readable registry with agentHint
├── decisions/ # Architectural decisions (ADRs)
│ └── ADR-NNNN-title.md
├── architecture/ # System structure docs
│ └── ARCH-NNNN-title.md
├── patterns/ # Reusable code patterns
│ └── PATTERN-NNNN-title.md
├── conventions/ # Coding rules
│ └── CONV-NNNN-title.md
├── features/ # Feature deep-dives
│ └── FEAT-NNNN-title.md
└── findings/ # Debug insights, gotchas
└── FINDING-NNNN-title.md
Knowledge Categories
| Category | ID Prefix | Purpose | When to Record |
|---|---|---|---|
| decision | ADR-NNNN |
Architectural choices with rationale | Non-trivial architectural decisions |
| architecture | ARCH-NNNN |
System structure and component relationships | Documenting system design |
| pattern | PATTERN-NNNN |
Reusable implementation approaches | New code patterns emerge |
| convention | CONV-NNNN |
Coding rules and constraints | Inconsistencies resolved |
| feature | FEAT-NNNN |
Feature-specific deep-dives | Complex feature needs documentation |
| finding | FINDING-NNNN |
Debug root causes, gotchas | Non-obvious bugs fixed |
File Format
Naming Convention
PREFIX-NNNN-short-kebab-title.md where PREFIX matches the category ID prefix.
Examples:
ADR-0001-nextjs-app-router.mdARCH-0001-system-overview.mdCONV-0001-named-exports.md
Frontmatter Schema
---
id: PREFIX-NNNN
title: Short descriptive title
status: proposed|accepted|deprecated|superseded
created: YYYY-MM-DD
updated: YYYY-MM-DD
tags: [tag1, tag2]
related: [ID1, ID2]
supersedes: ID (optional, ADRs only)
superseded-by: ID (optional, ADRs only)
---
Index Format
The docs/index.json file provides fast lookups:
{
"schemaVersion": "1.0.0",
"description": "Project documentation registry",
"updatedAt": "2026-02-28",
"categories": {
"decision": "Architectural choices and reasoning (ADRs)",
"architecture": "System structure, libs, data flow",
"pattern": "Reusable code patterns with examples",
"convention": "Coding rules and constraints",
"feature": "Deep-dive guides for specific features",
"finding": "Discovered gotchas and debug insights"
},
"entries": [{
"id": "ADR-0001",
"title": "Use Next.js App Router",
"category": "decision",
"status": "accepted",
"audience": ["agent", "human"],
"path": "docs/decisions/ADR-0001-nextjs-app-router.md",
"tags": ["nextjs", "routing"],
"agentHint": "check before choosing routing strategy or adding pages",
"related": ["ARCH-0001"]
}]
}
Key Fields
agentHint— Single sentence: when should an agent check this entry. Matched against current task keywords.audience—["agent", "human"]or["agent"]for agent-only entries.path— Relative to project root (e.g.,docs/decisions/ADR-0001.md).
Significance Threshold
Record when:
- Root cause took >10 minutes to find
- New pattern emerged others should follow
- Architectural decision could be questioned later
- Convention was inconsistent or newly established
- Research yielded key technology choice
Don't record:
- Trivial fixes (typos, imports, formatting)
- Well-known patterns (standard React hooks, basic CRUD)
- Information already in official docs
- Obvious bugs with simple fixes
Knowledge vs Agent Memory
| Aspect | Knowledge Base | Agent Memory |
|---|---|---|
| Scope | Team-wide, project-level | Agent-specific, session context |
| Persistence | Git-tracked, permanent | Auto-managed by Claude |
| Management | Explicit recording | Automatic |
| Purpose | Share learnings, preserve decisions | Task continuity, context |
| Format | Structured markdown + YAML | Unstructured notes |
Cross-Referencing
Entries reference each other by ID in related array:
related: [ADR-0001, PATTERN-005, FINDING-012]
This creates a knowledge graph linking related concepts.
Aspect Files
| File | Purpose |
|---|---|
adr-patterns.md |
ADR template and lifecycle |
knowledge-capture-guide.md |
When and how to capture knowledge |
sidecar-format-spec.md |
Data formats and schemas |
Workflow Integration
- During implementation: Discover patterns → capture to
patterns/ - During debugging: Find root cause → capture to
findings/ - During architecture: Make design choice → capture to
decisions/ - During review: Identify convention → capture to
conventions/ - During documentation: Document system structure → capture to
architecture/ - During feature work: Complex feature needs guide → capture to
features/
Sub-Skill Routing
When this skill is active and user intent matches a sub-skill, delegate:
| Intent | Sub-Skill | When |
|---|---|---|
| Capture knowledge | knowledge-capture |
After debug/plan, persist learning |
| Retrieve knowledge | knowledge-retrieval |
Before planning/research, search internal |
| Store data | data-store |
.epost-data/ directory patterns |
Related Skills
knowledge-retrieval— Search and retrieve knowledge entriesknowledge-capture— Post-task knowledge persistence workflow
References
references/adr-patterns.md— ADR template and lifecyclereferences/knowledge-capture-guide.md— Capture guidelinesreferences/sidecar-format-spec.md— Data format specifications
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?