Task Observer
Monitor task execution and capture skill improvement opportunities in real-time
What You Can Do
This skill acts as a persistent observer layer that monitors your task execution, tool interactions, and workflow decisions to surface skill improvement opportunities. It captures user corrections, effective patterns, and methodology insights during real-world work, then feeds these observations to your skill-creation process. Use it during any multi-step task, agentic workflow, or substantive work session to build an continuously improving skill library based on actual usage patterns.
Features
identifies repeatable workflows and decision-making approaches during task execution
captures user feedback and course corrections to surface methodology refinements
records tool sequences, dependencies, and handoff points worth preserving
spots moments where a custom skill would streamline repetitive work
triggers during feedback discussions to extract actionable skill observations
activates on skill-related keywords like 'skill observation', 'improvement opportunity', or 'One Skill to Rule Them All'
documents effective approaches and best practices discovered during work sessions
organizes observations within your existing skill structure for easy reference
Example Output
During a data analysis workflow:
- Observer notes: Used 3-step SQL → Python → visualization pipeline; user corrected chart formatting twice with specific color rules
- Captured pattern: SQL aggregation + Pandas transformation + Matplotlib styling = reusable data visualization skill
- Observation logged: Time spent on repetitive Pandas syntax could be reduced with custom transformation template
During API integration task:
- Observer detects: Authentication → validation → retry logic → error mapping pattern across 4 API calls
- Records: User refined error handling approach mid-task; original approach logged as anti-pattern
- Skill opportunity identified: Generic API error-handler skill would prevent this pattern in future integrations
Post-task review:
- Summary: 6 workflow optimization opportunities, 2 methodology refinements, 1 new skill candidate ready for creation
What's Included
- SKILL.md instruction file with activation triggers and behavioral guidelines:
- Observation log template for tracking patterns, corrections, and insights during work sessions:
- Skill opportunity checklist to evaluate whether captured patterns warrant new skill creation:
- Workflow analysis framework for extracting methodology from multi-step task execution:
- Integration guide for pairing with CLAUDE.md harness for automatic session-start activation:
Who It's For
- Software engineers & developers building reusable automation and skill libraries from workflow patterns
- Data scientists capturing effective analysis methodologies and tool-chain improvements
- DevOps engineers documenting repeatable deployment and infrastructure workflows
- Technical team leads creating institutional knowledge from team task execution patterns
- Prompt engineers & AI builders continuously improving their custom skill taxonomies based on real usage
Best For
- Multi-step technical workflows requiring tool coordination and decision-making
- Agentic tasks where Claude uses external APIs, databases, or file systems
- Iterative work sessions with user feedback and course corrections
- Complex processes that recur frequently and benefit from automation
- Building organization-wide skill libraries from team workflows
You might also like
You provide raw logs, metrics, and traces from your incident, and Claude systematically correlates them to identify root causes rapidly. Instead of drowning in noise, you get a focused investigation workflow that prioritizes the most likely failure points and guides you to actionable insights.
You can organize claim submissions by payer-specific requirements, analyze denial patterns from EOB data, draft professional appeals and payer correspondence, and create prior authorization narratives with clinical justification. This skill helps you systematize your entire claim lifecycle to reduce overhead by 30-40% and accelerate cash flow through faster denial resolution.
by rebelytics
Systematically investigate production incidents by correlating metrics, logs, and distributed traces to pinpoint root causes. You provide Claude with observability data from your infrastructure, and it builds a timeline, identifies anomalies, tests hypotheses, and delivers a structured incident report with actionable remediation steps. This accelerates MTTR and prevents recurrence through data-driven diagnostics.
by az9713
Rapidly analyze production incidents by correlating observability signals—logs, metrics, traces, and events—to identify the true root cause instead of symptoms. You'll reconstruct incident timelines, quantify impact scope, and generate both immediate remediation steps and long-term prevention measures.