# Fabric > AI-powered text processing using Fabric patterns with local Ollama Use when this capability is needed. - Skill: `tomevault-io/fabric-2` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/fabric-2` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/fabric-2/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/fabric-2 --- # Fabric Skill AI-powered text processing using [danielmiessler/fabric](https://github.com/danielmiessler/fabric) patterns with local Ollama inference. --- ## Overview | Aspect | Description | |--------|-------------| | Purpose | Run pre-built AI prompt patterns for logs, code review, commit messages | | Backend | Local Ollama (qwen2.5:32b primary, qwen2.5:7b-instruct fallback) | | Pattern | Capability Layering - thin wrapper over existing CLI scripts | | Cost | Free (local inference) | --- ## Quick Actions | Need | Action | Command | |------|--------|---------| | Analyze logs | AI analysis of Docker container logs | `/fabric:analyze-logs ` | | Commit message | Generate from staged changes | `/fabric:commit-msg` | | Code review | AI-powered code review | `/fabric:review-code ` | | List patterns | See all 234 available patterns | `/fabric patterns` | | Run any pattern | Execute arbitrary pattern | `/fabric run ` | --- ## Commands ### /fabric:analyze-logs Analyze Docker container logs for patterns, anomalies, and recommendations. ```bash /fabric:analyze-logs prometheus # Last 50 lines /fabric:analyze-logs nginx --lines 200 # More context /fabric:analyze-logs n8n --since 1h # Recent logs only ``` **Output sections**: Overview, Key Observations, Patterns/Anomalies, Recommendations ### /fabric:commit-msg Generate conventional commit messages from git diffs. ```bash /fabric:commit-msg # From staged changes /fabric:commit-msg --all # From all changes ``` **Output format**: `: ` with change list ### /fabric:review-code AI-powered code review with prioritized recommendations. ```bash /fabric:review-code src/server.ts # Single file /fabric:review-code --staged # Staged changes ``` **Output sections**: Overall Assessment, Prioritized Recommendations, Detailed Feedback ### /fabric patterns List all available Fabric patterns. ```bash /fabric patterns # List all 234 patterns /fabric patterns --search log # Filter by keyword ``` ### /fabric run Run any Fabric pattern directly. ```bash echo "text" | /fabric run extract_wisdom /fabric run summarize --file document.md ``` --- ## Architecture This skill follows the **Capability Layering Pattern**: ``` Layer 5: USER REQUEST "analyze the prometheus logs" ↓ Layer 4: PROMPT (this skill) Routes to correct script ↓ Layer 3: CLI Scripts/fabric-analyze-logs.sh prometheus ↓ Layer 2: CODE fabric-wrapper.sh → fabric CLI → Ollama ↓ Layer 1: INFRASTRUCTURE Ollama service (localhost:11434) ``` ### Scripts (Layer 3) | Script | Purpose | |--------|---------| | `Scripts/fabric-wrapper.sh` | Core wrapper with health checks, model fallback | | `Scripts/fabric-analyze-logs.sh` | Log analysis for Docker/files | | `Scripts/fabric-commit-msg.sh` | Commit message generation | | `Scripts/fabric-review-code.sh` | Code review | ### Features Built Into Scripts - **Ollama Health Check**: Auto-restarts Ollama if unresponsive - **Model Fallback**: 32b → 7b on timeout (configurable) - **Streaming Output**: Real-time results - **Error Handling**: Clear error messages --- ## Configuration ### Environment Variables | Variable | Default | Description | |----------|---------|-------------| | `FABRIC_PRIMARY_MODEL` | `qwen2.5:32b` | Primary model for analysis | | `FABRIC_FALLBACK_MODEL` | `qwen2.5:7b-instruct` | Fallback on timeout | | `FABRIC_TIMEOUT_PRIMARY` | `90` | Timeout in seconds for primary | | `FABRIC_TIMEOUT_FALLBACK` | `120` | Timeout for fallback | | `OLLAMA_URL` | `http://localhost:11434` | Ollama API endpoint | ### Model Selection Guide | Task | Recommended Model | Reason | |------|-------------------|--------| | Log analysis | 32b (default) | Benefits from deep reasoning | | Commit messages | 7b | Fast, simple task | | Code review | 7b (default) | Good balance of speed/quality | | Complex analysis | 32b | Use `--model 32b` flag | --- ## Patterns Reference Fabric includes 234 patterns. Most useful for this infrastructure: | Pattern | Use Case | Value | |---------|----------|-------| | `analyze_logs` | Docker/service troubleshooting | HIGH | | `summarize_git_diff` | Commit messages | HIGH | | `review_code` | Code review | HIGH | | `extract_wisdom` | Document summarization | MEDIUM | | `create_design_document` | Technical specs | MEDIUM | | `analyze_threat_report` | Security analysis | MEDIUM | Full pattern list: `fabric --list` or https://github.com/danielmiessler/fabric/tree/main/patterns --- ## Troubleshooting ### Ollama Not Responding The wrapper auto-restarts Ollama, but if issues persist: ```bash # Check Ollama status systemctl status ollama # Restart manually sudo systemctl restart ollama # Check models available ollama list ``` ### Model Timeout For large inputs, increase timeout: ```bash FABRIC_TIMEOUT_PRIMARY=180 fabric-analyze-logs.sh prometheus --lines 500 ``` Or force the faster model: ```bash fabric-analyze-logs.sh prometheus --model 7b ``` ### Pattern Not Found Update fabric patterns: ```bash fabric --update ``` --- ## Related - **Capability Layering Pattern**: @.claude/context/patterns/capability-layering-pattern.md - **Scripts Documentation**: @Scripts/README.md - **Ollama Management**: `/ollama` skill (if needed) - **Fabric GitHub**: https://github.com/danielmiessler/fabric --- ## Version History | Version | Date | Changes | |---------|------|---------| | 1.0.0 | 2026-01-22 | Initial skill creation with 4 commands | --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/davidmoneil) — claim your Tome and manage your conversions.