🔬 Repository Quality Improvement Report – Large File Decomposition Debt (2026-03-31)

[github-actions[bot]]

🎯 Repository Quality Improvement Report – Large File Decomposition Debt

Analysis Date: 2026-03-31
Focus Area: Large File Decomposition Debt (Custom)
Strategy Type: Custom – Repository-Specific
Custom Area: Yes – This focus area was selected because AGENTS.md explicitly states a hard limit of 300 lines for source files and a recommended split at 200 lines. Analysis reveals 174 files currently exceed the 300-line threshold, with the worst offenders reaching over 1,300 lines. This represents a significant and measurable technical debt unique to this codebase.

---

Executive Summary

The gh-aw codebase is healthy in many respects (test-to-source ratio >2:1, full lint/format enforcement, strong CI), but a significant pattern of oversized files has accumulated. The project's own AGENTS.md documentation sets a hard limit of 300 lines per file with a recommended target of 100–200 lines and a decision tree for splitting – yet 174 production source files exceed that hard limit, and 80 exceed 500 lines.

The largest violations are concentrated in pkg/cli/ (29 files >500 lines) and pkg/workflow/ (41 files >500 lines). The worst single file, gateway_logs.go at 1,332 lines, contains four distinct responsibility domains: RPC message parsing, guard policy handling, log aggregation, and rendering. pkg/constants/constants.go at 1,083 lines mixes semantic type definitions with constants for ten different subsystems.

Splitting these files into cohesive, single-responsibility units will improve navigability for new contributors, reduce merge conflict risk, make selective testing easier, and bring the codebase into alignment with its own stated guidelines.

---

Full Analysis Report

Focus Area: Large File Decomposition Debt

Current State Assessment

Metrics Collected:

Metric	Value	Status
Source files total	628	–
Average file size	~238 lines	✅
Files exceeding 300-line limit	174	❌
Files exceeding 500 lines	80	❌
Files exceeding 800 lines	17	❌
Largest file (gateway_logs.go)	1,332 lines	❌
pkg/workflow files >500 lines	41	❌
pkg/cli files >500 lines	29	❌
pkg/parser files >500 lines	7	❌

Strengths

• Average file size (238 lines) is within recommended range, suggesting the problem is concentrated rather than systemic
• All test files have correct //go:build tags
• Strong test-to-source ratio (>2:1 by LOC)
• Good use of file naming conventions – split files would be easy to name

Areas for Improvement

• 🔴 174 files exceed AGENTS.md hard limit of 300 lines (High)
• 🔴 pkg/cli/gateway_logs.go (1,332 lines) – four distinct domains in one file (High)
• 🔴 pkg/constants/constants.go (1,083 lines) – mixes type aliases and constants across 10+ subsystems (High)
• 🟡 pkg/cli/audit_report_render.go (1,045 lines) – rendering split across JSON and 20+ console section functions (Medium)
• 🟡 pkg/cli/logs_report.go (1,011 lines) – data builder and renderer combined (Medium)

Top 20 Files Exceeding Limits

Lines	File
1,332	pkg/cli/gateway_logs.go
1,083	pkg/constants/constants.go
1,045	pkg/cli/audit_report_render.go
1,011	pkg/cli/logs_report.go
1,007	pkg/cli/trial_command.go
987	pkg/workflow/checkout_manager.go
971	pkg/workflow/compiler_safe_outputs_config.go
971	pkg/workflow/cache.go
956	pkg/cli/logs_orchestrator.go
921	pkg/workflow/frontmatter_types.go
910	pkg/workflow/compiler_orchestrator_workflow.go
898	pkg/parser/remote_fetch.go
865	pkg/workflow/frontmatter_extraction_yaml.go
860	cmd/gh-aw/main.go
859	pkg/workflow/compiler_jobs.go
829	pkg/cli/pr_command.go
820	pkg/workflow/domains.go
814	pkg/parser/schema_suggestions.go
810	pkg/workflow/compiler_yaml_main_job.go
809	pkg/workflow/compiler_yaml.go

---

🤖 Tasks for Copilot Agent

NOTE TO PLANNER AGENT: The following 5 tasks are designed for GitHub Copilot coding agent execution. Each task is self-contained. Please split these into individual work items for processing.

---

Task 1: Split pkg/cli/gateway_logs.go into domain-cohesive files

Priority: High
Estimated Effort: Medium
Focus Area: Large File Decomposition

Description:
pkg/cli/gateway_logs.go (1,332 lines) contains four distinct responsibility domains:

1. Type definitions – GatewayLogEntry, GatewayMetrics, RPCMessageEntry and related structs
2. RPC message parsing – parseRPCMessages, buildToolCallsFromRPCMessages
3. Log aggregation & guard policy handling – parseGatewayLogs, calculateGatewayAggregates, buildGuardPolicySummary, isGuardPolicyErrorCode
4. Rendering – renderGatewayMetricsTable, displayAggregatedGatewayMetrics

Split this file into three cohesive files:

• pkg/cli/gateway_logs_types.go – all type definitions
• pkg/cli/gateway_logs_parser.go – parsing functions
• pkg/cli/gateway_logs_render.go – rendering and display functions
• Keep gateway_logs.go for orchestration/aggregation logic

Acceptance Criteria:

• File split into ≥3 files, each ≤300 lines
• All existing tests pass (go test ./pkg/cli/...)
• make fmt && make lint passes
• No logic changes – pure restructuring

Code Region: pkg/cli/gateway_logs.go

Split `pkg/cli/gateway_logs.go` (1,332 lines) into multiple cohesive files following single-responsibility principle. The file has four distinct domains:

1. Type definitions (GatewayLogEntry, GatewayMetrics, RPCMessageEntry, DifcFilteredEvent, GuardPolicyEvent, GatewayServerMetrics, GatewayToolMetrics, RPCMessageEntry, rpcRequestPayload, rpcToolCallParams, rpcResponsePayload, rpcError, rpcErrorData, rpcPendingRequest)
2. RPC message parsing (parseRPCMessages, findRPCMessagesPath, buildToolCallsFromRPCMessages)
3. Gateway log parsing and aggregation (parseGatewayLogs, processGatewayLogEntry, getOrCreateServer, getOrCreateTool, calculateGatewayAggregates, buildGuardPolicySummary, isGuardPolicyErrorCode, guardPolicyReasonFromCode, extractMCPToolUsageData, getSortedServerNames)
4. Rendering (renderGatewayMetricsTable, displayAggregatedGatewayMetrics, buildGuardPolicySummary)

Create these new files (all in package `cli`):
- `pkg/cli/gateway_logs_types.go` – all struct/type definitions
- `pkg/cli/gateway_logs_rpc.go` – RPC message parsing functions
- `pkg/cli/gateway_logs_render.go` – rendering and display functions
- Retain `pkg/cli/gateway_logs.go` for gateway log parsing, aggregation, and the main orchestration functions

Requirements:
- Pure refactoring only – no logic changes
- All files must be ≤400 lines
- Run `make fmt && make lint` after splitting
- Verify `go test ./pkg/cli/...` passes

---

Task 2: Split pkg/constants/constants.go into domain-specific files

Priority: High
Estimated Effort: Medium
Focus Area: Large File Decomposition

Description:
pkg/constants/constants.go (1,083 lines) contains constants for many unrelated subsystems. Split it into domain-specific files:

• pkg/constants/constants_types.go – semantic type aliases and their methods (LineLength, Version, FeatureFlag, URL, ModelName, JobName, StepID, etc.)
• pkg/constants/constants_engines.go – engine-specific constants (Copilot, Claude, Codex, Gemini, Playwright versions and defaults)
• pkg/constants/constants_sandbox.go – firewall/sandbox constants (AWF*, firewall paths, registries, containers)
• pkg/constants/constants_jobs.go – GitHub Actions job names, step IDs, job/step priority fields
• pkg/constants/constants.go – retain CLI prefix, workspace paths, and general constants

Acceptance Criteria:

• constants.go reduced to ≤300 lines
• Each new file ≤300 lines
• All existing tests pass (go test ./pkg/constants/...)
• make fmt && make lint passes

Code Region: pkg/constants/constants.go

Split `pkg/constants/constants.go` (1,083 lines) into domain-specific constant files. The file currently mixes semantic type definitions with constants from 10+ subsystems.

Proposed split:
1. `pkg/constants/constants_types.go` – Move all semantic type alias declarations and their String()/IsValid() method implementations (LineLength, Version, FeatureFlag, URL, ModelName, JobName, StepID, CommandPrefix, WorkflowID, EngineName, DocURL, GitHubToolName, GitHubToolset, GitHubAllowedTools, GitHubToolsets)
2. `pkg/constants/constants_engines.go` – Engine version defaults (DefaultCopilotVersion, DefaultClaudeCodeVersion, DefaultCodexVersion, DefaultGeminiVersion, DefaultPlaywrightMCPVersion, DefaultQmdVersion, DefaultMCPSDKVersion, DefaultGitHubScriptVersion, engine feature flags, model names)
3. `pkg/constants/constants_sandbox.go` – Firewall and sandbox constants (AWF*, DefaultFirewallVersion, DefaultFirewallRegistry, DefaultMCPGatewayVersion, DefaultMCPGatewayContainer, DefaultMCPGatewayPayloadDir, DefaultMCPGatewayPayloadSizeThreshold, mount constants)
4. `pkg/constants/constants_jobs.go` – GitHub Actions job/step identifiers (JobName constants, StepID constants, PriorityStepFields, PriorityJobFields, PriorityWorkflowFields, SharedWorkflowForbiddenFields, IgnoredFrontmatterFields)
5. `pkg/constants/constants.go` – Retain: CLIExtensionPrefix, GhAwRootDir, GhAwRootDirShell, runtime version defaults (Node, Python, Ruby, Go), general utility vars (CopilotStemCommands, DefaultBashTools, DefaultAllowedMemoryExtensions), and GetWorkflowDir()

All files must remain in `package constants`. Pure refactoring only – no logic changes.
Run `make fmt && make lint` and verify `go test ./pkg/constants/...` passes after splitting.

---

Task 3: Split pkg/cli/audit_report_render.go into render-domain files

Priority: High
Estimated Effort: Medium
Focus Area: Large File Decomposition

Description:
pkg/cli/audit_report_render.go (1,045 lines) contains ~25 rendering functions across JSON output, console rendering, and 20+ section-specific renderers (overview, metrics, tool usage, MCP, guard policies, firewall, etc.). These can be cleanly split by rendering target and domain:

• pkg/cli/audit_render_json.go – renderJSON
• pkg/cli/audit_render_overview.go – high-level overview sections (renderConsole, renderOverview, renderMetrics, renderEngineConfig)
• pkg/cli/audit_render_security.go – security-domain sections (renderGuardPolicySummary, renderFirewallAnalysis, renderRedactedDomainsAnalysis, renderPolicyAnalysis)
• pkg/cli/audit_render_tools.go – tool usage sections (renderToolUsageTable, renderMCPToolUsageTable, renderMCPServerHealth)
• pkg/cli/audit_report_render.go – keep remaining sections and shared helpers

Acceptance Criteria:

• Original file reduced to ≤400 lines
• Each new file ≤300 lines
• make fmt && make lint passes
• go test ./pkg/cli/... passes

Code Region: pkg/cli/audit_report_render.go

Split `pkg/cli/audit_report_render.go` (1,045 lines) by rendering domain. This file has ~25 functions covering JSON rendering, general console rendering, and 20+ section-specific console renderers.

Proposed split into files in `package cli`:
1. `pkg/cli/audit_render_json.go` – renderJSON function and any JSON-specific helpers
2. `pkg/cli/audit_render_security.go` – security domain: renderGuardPolicySummary, renderFirewallAnalysis, renderRedactedDomainsAnalysis, renderPolicyAnalysis
3. `pkg/cli/audit_render_tools.go` – tool usage domain: renderToolUsageTable, renderMCPToolUsageTable, renderMCPServerHealth, renderSafeOutputSummary
4. `pkg/cli/audit_render_analysis.go` – analysis domain: renderTaskDomain, renderBehaviorFingerprint, renderAgenticAssessments, renderSessionAnalysis, renderPromptAnalysis, renderKeyFindings, renderRecommendations
5. `pkg/cli/audit_report_render.go` – keep: renderConsole (main entry point), renderOverview, renderMetrics, renderJobsTable, renderCreatedItemsTable, renderPerformanceMetrics, renderEngineConfig, renderAuditComparison, formatUnixTimestamp

Pure refactoring only. Run `make fmt && make lint` and `go test ./pkg/cli/...` after splitting.

---

Task 4: Split pkg/cli/logs_report.go into builder and renderer files

Priority: Medium
Estimated Effort: Small
Focus Area: Large File Decomposition

Description:
pkg/cli/logs_report.go (1,011 lines) contains two distinct concerns that should be separated:

1. Data building – buildLogsData, buildToolUsageSummary, buildMissingToolsSummary, buildMCPFailuresSummary, buildAccessLogSummary, buildFirewallLogSummary, buildRedactedDomainsSummary, buildMCPToolUsageSummary, buildCombinedErrorsSummary
2. Rendering – renderLogsJSON, renderLogsConsole, writeSummaryFile

Acceptance Criteria:

• logs_report.go reduced to ≤500 lines
• New file ≤400 lines
• make fmt && make lint passes
• go test ./pkg/cli/... passes

Code Region: pkg/cli/logs_report.go

Split `pkg/cli/logs_report.go` (1,011 lines) to separate data aggregation from rendering.

Move rendering functions to `pkg/cli/logs_report_render.go` (package `cli`):
- renderLogsJSON
- renderLogsConsole
- writeSummaryFile

Keep in `pkg/cli/logs_report.go`:
- buildLogsData
- isValidToolName
- buildToolUsageSummary
- addUniqueWorkflow
- aggregateSummaryItems
- buildMissingToolsSummary
- buildMissingDataSummary
- buildMCPFailuresSummary
- aggregateDomainStats
- convertDomainsToSortedSlices
- buildAccessLogSummary
- buildFirewallLogSummary
- buildRedactedDomainsSummary
- buildMCPToolUsageSummary
- buildCombinedErrorsSummary

Pure refactoring only. Run `make fmt && make lint` and `go test ./pkg/cli/...` after splitting.

---

Task 5: Add file-size enforcement to CI

Priority: Medium
Estimated Effort: Small
Focus Area: Large File Decomposition

Description:
The 300-line hard limit is documented in AGENTS.md but not enforced by CI. New files are added without automated checks, allowing the debt to grow silently. Adding a simple script that warns (or fails) when a newly introduced file exceeds the limit will prevent regression.

Add a scripts/check-file-sizes.sh script that:

1. Lists all non-test .go files exceeding 500 lines (more lenient threshold to avoid immediate failures from existing violations)
2. Compares against a baseline allowlist of current violators
3. Fails CI if new files exceed 500 lines that are not in the allowlist
4. Prints a helpful message referencing AGENTS.md guidelines

Integrate this check into the Makefile as make check-file-sizes and optionally add it to the CI workflow.

Acceptance Criteria:

• scripts/check-file-sizes.sh exists and is executable
• make check-file-sizes target added to Makefile
• Script generates a baseline file listing current violators
• Script fails (exit code 1) when new files exceed the threshold
• Script prints actionable message referencing AGENTS.md

Code Region: scripts/, Makefile

Add a file-size enforcement script to prevent new oversized files from being added to the codebase.

1. Create `scripts/check-file-sizes.sh`:
   - Accept a threshold parameter (default: 500 lines for CI, matching hard limit philosophy)
   - Find all non-test .go files exceeding the threshold in pkg/ and cmd/
   - Compare against an allowlist file `scripts/file-size-allowlist.txt` (existing violators are excluded)
   - Exit 0 if only known violators are over threshold, exit 1 if new files are found
   - Print clear messages: "New oversized file detected: <file> (<lines> lines). See AGENTS.md for file splitting guidelines."
   - Generate allowlist: running with `--update-allowlist` flag updates `scripts/file-size-allowlist.txt`

2. Populate `scripts/file-size-allowlist.txt` with all current files >500 lines (run the script with --update-allowlist to generate this)

3. Add to `Makefile`:
   ```makefile
   check-file-sizes: ## Check for new oversized files (>500 lines)
       @./scripts/check-file-sizes.sh

4. Make the script executable: chmod +x scripts/check-file-sizes.sh

Run make check-file-sizes to verify the script works. The script should succeed (exit 0) since all current violations are known.

---

## 📊 Historical Context

<details>
<summary><b>Previous Focus Areas</b></summary>

| Date | Focus Area | Type | Custom | Key Outcomes |
|------|------------|------|--------|--------------|
| 2026-03-31 | Large File Decomposition Debt | Custom | Yes | First run – identified 174 files exceeding 300-line guideline; 5 tasks generated |

</details>

---

## 🎯 Recommendations

### Immediate Actions (This Week)
1. **Split `gateway_logs.go`** (1,332 lines → 4 files) – Priority: High
2. **Split `constants.go`** (1,083 lines → 5 files) – Priority: High
3. **Split `audit_report_render.go`** (1,045 lines → 5 files) – Priority: High

### Short-term Actions (This Month)
1. **Split `logs_report.go`** (1,011 lines → 2 files) – Priority: Medium
2. **Add file-size CI enforcement** to prevent regression – Priority: Medium

### Long-term Actions (This Quarter)
1. Systematically address the remaining 169 files exceeding 300 lines, prioritizing `pkg/workflow/` (41 files >500 lines)
2. Consider adding a `golangci-lint` custom rule or `revive` configuration for max file length

---

## 📈 Success Metrics

Track these metrics to measure improvement in **Large File Decomposition**:

- **Files >300 lines**: 174 → Target: <50 (within 1 quarter)
- **Files >500 lines**: 80 → Target: <20 (within 1 quarter)
- **Largest single file**: 1,332 → Target: <400 lines
- **`pkg/workflow` files >500 lines**: 41 → Target: <15
- **`pkg/cli` files >500 lines**: 29 → Target: <10

---

## Next Steps

1. Review and prioritize the 5 tasks above
2. Assign tasks 1–3 to Copilot coding agent via planner agent (high priority)
3. Assign tasks 4–5 for follow-on sprint
4. Re-evaluate this focus area after 4 weeks to track metrics progress

---

**References:**
- [§23800005791](https://github.com/github/gh-aw/actions/runs/23800005791)


> AI generated by [Repository Quality Improvement Agent](https://github.com/github/gh-aw/actions/runs/23800005791) · [history](https://github.com/search?q=repo%3Agithub%2Fgh-aw+%22gh-aw-workflow-call-id%3A+github%2Fgh-aw%2Frepository-quality-improver%22&type=discussions)
> - [x] expires <!-- gh-aw-expires: 2026-04-01T13:39:34.250Z --> on Apr 1, 2026, 1:39 PM UTC

<!-- gh-aw-workflow-id: repository-quality-improver -->
<!-- gh-aw-workflow-call-id: github/gh-aw/repository-quality-improver -->

[github-actions[bot]]

This discussion has been marked as outdated by Repository Quality Improvement Agent.

A newer discussion is available at Discussion #23903.