levnikolaevich/ln-622-build-delivery-gate-auditor

Paths: File paths (references/, ../ln-*) are relative to this skill directory.

Build & Delivery Gate Auditor (L3 Worker)

Type: L3 Worker

Specialized worker auditing executable delivery gates and CI/build feedback.

Purpose & Scope

• Worker in codebase audit pipeline
• Audit codebase for build and delivery gate issues (Category 2: Critical Priority)
• Check compiler/linter/type/test/CI failures and build config that blocks or weakens delivery
• Emit FIX_DELIVERY_GATE, FAIL_CI_ON_SIGNAL, or REMOVE_STALE_SKIP
• Return structured findings with severity, location, effort, actions, and recommendations
• Calculate compliance score (X/10) for Build Health category

Inputs

MANDATORY READ: Load references/audit_worker_core_contract.md and references/mcp_tool_preferences.md. Tool policy: You may run as an isolated subagent where host AGENTS.md is not in scope, so default to hex-line MCP (read_file, grep_search, outline) over built-in Read/Grep when inspecting source. Load references/mcp_integration_patterns.md only when MCP behavior is unclear.

Receives contextStore with: tech_stack (including build_tool, test_framework), best_practices, principles, codebase_root, output_dir.

Workflow

Detection policy: use two-layer detection (candidate scan, then context verification); load references/two_layer_detection.md only when the verification method is ambiguous.

1. Parse Context: Extract tech stack, build tools, test framework, output_dir from contextStore
2. Run Build Checks (Layer 1): Execute compiler, linter, type checker, tests (see Audit Rules below)
3. Analyze Output Context (Layer 2): For deprecation warnings -- read notice to determine if removal is imminent or distant. For config issues -- check if dev-only or production config.
4. Collect Findings: Record each violation with severity, location, effort, recommendation
5. Calculate Score: Count violations by severity, calculate compliance score (X/10)
6. Write Report: Build full markdown report in memory per references/templates/audit_worker_report_template.md, write to {output_dir}/ln-622--global.md in single Write call
7. Trend Tracking: Append build_health metric to results_log per references/results_log_pattern.md. Metric: build_health | 0-10 | penalty formula. Calculate delta and status (improving/stable/declining) vs previous run.
8. Return Summary: Return minimal summary (see Output Format)

Audit Rules (Priority: CRITICAL)

MANDATORY READ: Load references/ci_tool_detection.md for build commands, linter commands, type checker commands, and test framework commands per ecosystem.

1. Compiler/Linter Errors

What: Syntax errors, compilation failures, linter rule violations

Detection: Use ci_tool_detection.md Command Registry (Build + Linters sections). Check exit code, parse stderr for errors. Use JSON output flags where available.

Linters: Use ci_tool_detection.md Linters table. Use --format json / --output-format json for structured output.

Severity:

• CRITICAL: Compilation fails, cannot build project
• HIGH: Linter errors (not warnings)
• MEDIUM: Linter warnings
• LOW: Stylistic linter warnings (formatting)

Recommendation: Fix errors before proceeding, configure linter rules, add pre-commit hooks

Effort: S-M (fix syntax error vs refactor code structure)

2. Unsupported API Warnings

What: Usage of unsupported APIs, libraries, or language features

Detection:

• Compiler warnings: UnsupportedApiWarning, @obsolete in stack trace
• Dependency warnings: npm outdated, pip list --outdated
• Static analysis: Grep for @obsolete annotations

Severity:

• CRITICAL: Unsupported API removed in next major version (imminent breakage)
• HIGH: Unsupported with migration path available
• MEDIUM: Unsupported but still supported for 1+ year
• LOW: Soft deprecation (no removal timeline)

Recommendation: Migrate to recommended API, update dependencies, refactor code

Effort: M-L (depends on API complexity and usage frequency)

3. Type Errors

What: Type mismatches, missing type annotations, type checker failures

Detection: Use ci_tool_detection.md Command Registry (Type Checkers section).

Severity:

• CRITICAL: Type error prevents compilation (tsc fails, cargo check fails)
• HIGH: Runtime type error likely (implicit any, missing type guards)
• MEDIUM: Missing type annotations (code works but untyped)
• LOW: Overly permissive types (any, unknown without narrowing)

Recommendation: Add type annotations, enable strict mode, use type guards

Effort: S-M (add types to single file vs refactor entire module)

4. Failed or Skipped Test Gates

What: Test command failures, stale skip markers, or CI configuration that allows known failures to pass

Detection: Use ci_tool_detection.md Command Registry (Test Frameworks section). Use JSON output flags for structured parsing.

Severity:

• CRITICAL: Test failures in CI/production code
• HIGH: Skipped tests for critical features (payment, auth)
• MEDIUM: Skipped tests for non-critical features
• LOW: Skipped tests with "TODO" comment (acknowledged debt)

Recommendation: Fix failing tests, remove stale skip markers, or make CI fail on this signal. Do not audit test portfolio value here.

Effort: S-M (update test assertion vs redesign test strategy)

5. Build Configuration Issues

What: Misconfigured build tools, missing scripts, incorrect paths

Detection:

• Missing build scripts in package.json, Makefile, build.gradle
• Incorrect paths in tsconfig.json, webpack.config.js, Cargo.toml
• Missing environment-specific configs (dev, staging, prod)
• Unused or conflicting build dependencies

Severity:

• CRITICAL: Build fails due to misconfiguration
• HIGH: Build succeeds but outputs incorrect artifacts (wrong target, missing assets)
• MEDIUM: Suboptimal config (no minification, missing source maps)
• LOW: Unused config options

Recommendation: Fix config paths, add missing build scripts, optimize build settings

Effort: S-M (update config file vs redesign build pipeline)

Scoring Algorithm

MANDATORY READ: Load references/audit_scoring.md.

Output Format

MANDATORY READ: Load references/templates/audit_worker_report_template.md.

Write JSON summary per references/audit_summary_contract.md. In managed mode the caller passes both runId and summaryArtifactPath; in standalone mode the worker generates its own run-scoped artifact path per shared contract.

Write report to {output_dir}/ln-622--global.md with category: "Build & Delivery Gate" and checks: compilation_errors, linter_warnings, type_errors, test_gate_failures, build_config.

Return summary per references/audit_summary_contract.md.

When summaryArtifactPath is absent, write the standalone runtime summary under .hex-skills/runtime-artifacts/runs/{run_id}/evaluation-worker/{worker}--{identifier}.json and optionally echo the same summary in structured output.

Report written: .hex-skills/runtime-artifacts/runs/{run_id}/audit-report/ln-622--global.md
Score: X.X/10 | Issues: N (C:N H:N M:N L:N)

Critical Rules

Apply the already-loaded references/audit_worker_core_contract.md.

• Do not auto-fix: Report violations only
• Tech stack aware: Use contextStore to run appropriate build commands (npm vs cargo vs gradle)
• Exit code checking: Always check exit code (0 = success, non-zero = failure)
• Timeout handling: Set timeout for build/test commands (default 5 minutes)
• Environment aware: Run in CI mode if detected (no interactive prompts)
• Unique angle: Audit only executable delivery gates. Do not audit test portfolio quality or dependency freshness.
• Action required: Every finding uses FIX_DELIVERY_GATE, FAIL_CI_ON_SIGNAL, or REMOVE_STALE_SKIP.

Monitor (2.1.98+): For build/lint/typecheck/test commands expected >30s, use Monitor. Fallback: Bash(run_in_background=true).

Definition of Done

Apply the already-loaded references/audit_worker_core_contract.md.

• [ ] contextStore parsed successfully (including output_dir)
• [ ] All 5 delivery gate checks completed (compiler, linter, type checker, test gate, config)
• [ ] Findings collected with severity, location, effort, action, recommendation
• [ ] Score calculated using penalty algorithm
• [ ] Report written to {output_dir}/ln-622--global.md (atomic single Write call)
• [ ] build_health metric appended to results_log with trend status
• [ ] Summary written per contract

Reference Files

• Audit output schema: references/audit_output_schema.md
• CI tool detection: references/ci_tool_detection.md
• Results log pattern: references/results_log_pattern.md
• Build audit rules: references/build_rules.md

---

Version: 3.0.0 Last Updated: 2025-12-23