leverage-point-audit

Leverage Point Audit

Audit a codebase against the 12 leverage points framework to identify gaps and improve agentic coding success.

When to Use

Before starting a new agentic coding project
When agents are failing or requiring many attempts
When KPIs (Size, Attempts, Streak, Presence) are not improving
For periodic health checks of agentic capability

The 12 Leverage Points

In-Agent (Core Four)

Context - CLAUDE.md, README, project docs
Model - Appropriate model selection
Prompt - Clear instructions and templates
Tools - Required capabilities available

Through-Agent (External)

Standard Out - Logging for visibility
Types - Information Dense Keywords (IDKs)
Documentation - Agent-specific context
Tests - Self-correction capability (HIGHEST LEVERAGE)
Architecture - Navigable codebase structure
Plans - Meta-work communication
Templates - Reusable prompts (slash commands)
ADWs - Autonomous workflows

Audit Workflow

Step 1: Check Context (Leverage Points 1-4)

CLAUDE.md presence:

Search for: CLAUDE.md, .claude/CLAUDE.md Check: Does it explain the project? Conventions? Common commands?

README.md quality:

Search for: README.md Check: Does it explain structure? How to run? How to test?

Permissions configuration:

Search for: .claude/settings.json Check: Are required tools allowed?

Step 2: Check Visibility (Leverage Point 5)

Standard out patterns:

Search for: print(, console.log(, logger., logging. Check: Are success AND error cases logged? Check: Can agent see what's happening?

Anti-pattern detection:

Look for: Silent returns, bare except blocks, empty catch blocks These prevent agent visibility.

Step 3: Check Searchability (Leverage Point 6)

Type definitions:

Search for: interface, type, class, BaseModel, dataclass Check: Are names information-dense? (Good: UserAuthToken, Bad: Data)

Step 4: Check Documentation (Leverage Point 7)

Internal docs:

Search for: *.md files, docstrings, comments Check: Do they explain WHY, not just WHAT?

Step 5: Check Validation (Leverage Point 8) - HIGHEST PRIORITY

Test presence:

Search for: test_*.py, *.test.ts, *.spec.ts, *_test.go Check: Do tests exist? Are they comprehensive?

Test commands:

Check: Is there a simple test command? (npm test, pytest, etc.) Check: Do tests run quickly?

Step 6: Check Architecture (Leverage Point 9)

Entry points:

Check: Are entry points obvious? (main.py, index.ts, server.py)

File organization:

Check: Consistent structure? Related files grouped? Check: File sizes reasonable? (< 1000 lines)

Step 7: Check Templates (Leverage Point 11)

Slash commands:

Search for: .claude/commands/ Check: Are common workflows automated?

Step 8: Check ADWs (Leverage Point 12)

Automation:

Search for: GitHub Actions, hooks, triggers Check: Are workflows automated?

Output Format

After audit, provide:

Summary Table

Leverage Point Status Priority Recommendation

Context Good/Fair/Poor High/Med/Low Specific action

... ... ... ...

Priority Actions

List top 3-5 improvements in order of impact:

[Highest Impact] - Specific recommendation
[High Impact] - Specific recommendation
[Medium Impact] - Specific recommendation

Detailed Findings

For each leverage point:

Current state
Specific gaps found
Recommended improvements
Example of what good looks like

Example Audit Output