---

name: code-inspection description: >- Audit newly written code against the Implementation Blueprint and context/coding-standards.md. Execute this after building a feature to verify production readiness. Do NOT execute this to automatically fix bugs; this skill only reports issues for the developer to decide on. version: 1.0.0 verified-on: [cline]

Code Inspection

0. Identity

• Role: Quality Assurance Lead. Verifies that implementation matches the blueprint and respects global architectural boundaries.
• Authority: Owns the post-build verification step. Cannot alter code directly during inspection.
• Must not define: The rules themselves.
• Normative base: The Implementation Blueprint, context/coding-standards.md, and context/architecture-blueprint.md.

1. Intent (9 Dimensions)

| # | Dimension | Value | | --- | ---------------- | ------------------------------------------------------------------- | | 1 | Task | Audit completed code against requirements and rules. | | 2 | Target Tool | Any agent runtime. | | 3 | Output Format | Three-layered inspection report (Requirements, System, Production). | | 4 | Constraints | Do not execute fixes. Output the report and stop. | | 5 | Input | Completed feature code. | | 6 | Context | Prevents untested, non-compliant code from entering production. | | 7 | Audience | The human developer. | | 8 | Success Criteria | A clear, prioritized report is presented to the developer. | | 9 | Examples | See §10. |

2. Trigger Matrix

| Trigger | Fire? | Notes | | -------------------------------------- | ----- | ------------------------------------- | | "Run inspect", "Review this feature" | YES | Core trigger. | | "Check if I followed the architecture" | YES | Core trigger. | | "Fix all the bugs in this file" | NO | Belongs to triage/standard execution. |

3. Execution Workflow

Step 1: Context Loading

• Action: Read the Implementation Blueprint for this feature and context/coding-standards.md.
• Input: Workspace files.
• Stop Condition: If blueprint cannot be found, ask the developer for the success criteria.
• Validation: Baseline expectations are established.

Step 2: Layer 1 - Blueprint Alignment

• Action: Compare the written code to the execution steps in the Blueprint. Identify missing requirements or unauthorized scope creep.
• Input: Source code vs Blueprint.
• Stop Condition: None.
• Validation: Feature completeness is verified.

Step 3: Layer 2 - System Integrity

• Action: Check code against context/coding-standards.md and context/architecture-blueprint.md. Verify boundary conditions (e.g., no database calls in UI components).
• Input: Source code vs Rules.
• Stop Condition: None.
• Validation: Architectural boundaries are respected.

Step 4: Layer 3 - Production Readiness

• Action: Scan for missing error handling, unhandled edge cases, and hardcoded secrets.
• Input: Source code.
• Stop Condition: None.
• Validation: Production stability is verified.

Step 5: Report Generation

• Action: Output the findings categorized by severity (Critical, Important, Minor). Stop and wait for the developer to command the next action.
• Input: Audit results.
• Stop Condition: Halt entirely after printing the report.
• Validation: Developer receives actionable intelligence.

4. Output Specification

# Inspection Report

## Layer 1: Blueprint Alignment

- [PASS/FAIL] [Details]

## Layer 2: System Integrity

- [PASS/FAIL] [Details]

## Layer 3: Production Readiness

- [PASS/FAIL] [Details]

## Prioritized Issues

- **Critical:** [Issue]
- **Minor:** [Issue]

Awaiting your decision on what to fix.

5. Validation Gate

• [ ] All three layers (Blueprint, System, Production) were checked.
• [ ] Issues were categorized by severity.
• [ ] No code was modified during this session.
• [ ] Agent halted completely after presenting the report.

6. Anti-Triggers and Calibration

• Under-execution threshold: Providing a generic "Looks good!" without checking the specific rules.
• Over-execution threshold: Automatically fixing the bugs found without developer permission.

7. Anti-Pattern Compliance

| Step | Prevents AP | Mechanism | | ---- | ---------------------------- | ------------------------------------------- | | 1 | AP-9 (no verification) | Establishes explicit verification criteria. | | 3 | AP-27 (no stack constraints) | Enforces boundaries against the rules file. | | 5 | AP-45 (no human review) | Forces human decision on what to fix. | | 5 | AP-51 (silent token drain) | Stops execution completely after reporting. |

8. Versioning & Changelog

• Version: 1.0.0
• Changelog:
  • 1.0.0 — Initial enterprise tier implementation.

9. Portability Matrix

| Runtime | Status | | ------- | -------- | | Cline | verified |

10. Examples

Input: "Inspect the new auth route." Output: Agent reads rules, finds a missing try/catch block (Layer 3), finds a DB call in the wrong directory (Layer 2), and presents the report for the developer to triage.