01
Using Skills
Master Claude's skills first—check before any action to ensure disciplined, effective workflows.
Intelligence Scale
Empowers the AI-driven development.
Install in Claude Code
1
/plugin marketplace add WeZZard/skills 2
/plugin install intelligence-scale@wezzard-skills Design Philosophy
Session Start
User Prompt Submit
Pre Tool Use
Post Tool Use
Post Tool Use Failure
Exit Plan Mode
Enter Plan Mode
Stop
Subagent Stop
Subagent Spawn
Write Plan
Ensures plans follow a reviewable, consistent structure
Triggers structured plan template when writing the plan file
Plan Writing Harness
Prevents ad-hoc planning by requiring template compliance
Enforces use of write-plan skill before writing plan files
Parallelism Reminder
Maximizes concurrent execution of independent work
Reminds to spawn subagents for independent tasks
Parallelism Reminder
Maximizes concurrent execution of independent work
Reminds to spawn subagents for independent tasks
Check Plan File
Measures the gap between plan execution and the goals in the plan file.
Checks the plan file after plan execution.
Recovery Skill Prompt
Prevents goal drift by prompting plan re-alignment
Suggests recover-from-errors skill on tool failures
Explore
Emphasizes web search during exploration, injecting a chain-of-thought that checks ground truth source dates and resolves conflicts by recency.
Raw
Codebase only
→
Intelligence Scale
Explore the web and codebase
Plan
Agent-first and human-first-oriented plan structure design.
Raw
Freeform
→
Intelligence Scale
Before/after comparison. Explicit diagrams for algorithms and workflow.
Task
Every user prompt triggers a reminder to spawn subagents for independent tasks, ensuring maximum throughput without manual orchestration.
Plan-File Based Audition
After execution, every task is verified against the plan with evidence-based status reporting: Done, Partial, or Missing.
Beyond Error Resilience
When tools fail, the natural instinct is to generalize or guess. Intelligence Scale prompts re-alignment with the plan, preventing cascading errors.
Raw
Tool fails → guess new params → drift from goal → waste context
→
Intelligence Scale
Tool fails → check plan → verify alignment → fix specific issue
Explore
Emphasizes web search during exploration, injecting a chain-of-thought that checks ground truth source dates and resolves conflicts by recency.
Raw
Codebase only
↓
Intelligence Scale
Explore the web and codebase
Plan
Agent-first and human-first-oriented plan structure design.
Raw
Freeform
↓
Intelligence Scale
Before/after comparison. Explicit diagrams for algorithms and workflow.
Task
Every user prompt triggers a reminder to spawn subagents for independent tasks, ensuring maximum throughput without manual orchestration.
Plan-File Based Audition
After execution, every task is verified against the plan with evidence-based status reporting: Done, Partial, or Missing.
Beyond Error Resilience
When tools fail, the natural instinct is to generalize or guess. Intelligence Scale prompts re-alignment with the plan, preventing cascading errors.
Raw
Tool fails → guess new params → drift from goal → waste context
↓
Intelligence Scale
Tool fails → check plan → verify alignment → fix specific issue
Skills
6 available 02
Brainstorming
Turn ideas into fully formed designs through collaborative dialogue
03
Write Plan
Create structured development plans with built-in verification and human oversight
04
Execute Plan
Turn your development plans into working code with automated execution and human oversight.
05
Audit Plan
Verify every planned change was implemented correctly with evidence-based auditing
06
Recover From Errors
Stay on track when tools fail by re-aligning with your plan instead of guessing.