codebase-audit

name	codebase-audit
description	Scan the dashboard-agent-swarm codebase for stubs, TODOs, hardcoded values, broken routes, and unimplemented features. Produces a prioritized fix list. Run this after any major feature branch merge to keep technical debt visible.
disable-model-invocation	true

Codebase Audit — Dashboard Agent Swarm

What This Skill Does

Launches three parallel sub-agents to audit the codebase from different angles, then produces a prioritized fix list with file:line references.

Pre-flight

Verify you are in the dashboard-agent-swarm directory or the archonx-os-main root:

ls src/pages/ && echo "✓ in dashboard-agent-swarm" || ls dashboard-agent-swarm/src/pages/ && echo "✓ in root"

---

Phase 1: Parallel Research (3 sub-agents)

Launch all three simultaneously using the Task tool.

Sub-agent 1: Stub & TODO Hunter

Scan every file in src/ and server/ for the following patterns. Return a table with: file path, line number, pattern type, and the exact code snippet.

Patterns to find:

1. // TODO // FIXME // HACK // XXX comments
2. setTimeout used as a fake async operation (not for UI animation)
3. Hardcoded strings that should come from env: API keys, URLs, org IDs, model names
4. console.log statements that should be proper logging
5. disabled props on buttons with no issue tracker reference
6. Empty state placeholders with no data fetching wired up
7. Promise.resolve(hardcodedData) — mock service functions
8. // Mock or // Stub or // placeholder comments
9. Route handlers that always return the same static response
10. any TypeScript type where a proper type should exist

Sort results by severity: CRITICAL (broken functionality) > HIGH (missing feature) > MEDIUM (tech debt) > LOW (cosmetic).

Sub-agent 2: Dead Code & Unused Imports

Scan every file in src/ for:

1. Imported packages that are never used (@tanstack/react-query is installed — find any useQuery/useMutation calls; if none exist, flag the package as unused)
2. React components defined but never imported anywhere
3. TypeScript types/interfaces defined but never referenced
4. Functions defined in services/ that are never called
5. Routes registered in src/App.tsx that have no corresponding nav item in AppSidebar.tsx
6. Nav items in AppSidebar.tsx that have no route in App.tsx
7. Server routes mounted in server/index.ts that have no corresponding client fetch in src/

Return a deduplicated list with file paths.

Sub-agent 3: Broken Wiring & Runtime Risks

Analyze the codebase for runtime issues:

1. Fetch calls to hardcoded ports/URLs that will fail in production (e.g., http://localhost:50001)
2. Missing error boundaries around pages that do network calls
3. useEffect with missing or incorrect dependency arrays
4. State that is fetched but never displayed
5. Forms with onSubmit that do nothing or just call preventDefault
6. Proxy targets in vite.config.ts that point to services not documented anywhere
7. Environment variables read in server code with no .env.example entry
8. Missing TypeScript return types on exported functions

Return a prioritized list with file paths and line numbers.

Wait for all three sub-agents to complete before proceeding.

---

Phase 2: Build Fix List

Merge findings from all three sub-agents. Deduplicate. Sort by:

1. CRITICAL — broken functionality (server returns mock, button is disabled with no path forward)
2. HIGH — missing live data (hardcoded status, unused React Query)
3. MEDIUM — tech debt (console.log, any types, missing error handling)
4. LOW — cosmetic or minor (unused imports, missing return types)

Create a TodoWrite list with one item per CRITICAL and HIGH finding.

---

Phase 3: Auto-Fix Pass

For each CRITICAL item:

1. Read the file
2. Implement the fix (wire up real data, remove hardcoded values)
3. Verify TypeScript compiles: npx tsc --noEmit
4. Mark the todo as completed

For HIGH items — implement fixes where straightforward (< 20 lines), flag the rest for manual review.

Skip MEDIUM and LOW items in the auto-fix pass — document them in the output report instead.

---

Phase 4: Report

Output a structured report:

## Codebase Audit — <date>

### Summary
- Files scanned: [N]
- Issues found: [N] (CRITICAL: N, HIGH: N, MEDIUM: N, LOW: N)
- Issues auto-fixed: [N]
- Issues requiring manual attention: [N]

### CRITICAL Issues Fixed
- [file:line] — description — fix applied

### HIGH Issues (Fixed)
- [file:line] — description — fix applied

### Remaining Issues (Manual Review Required)
- MEDIUM: [file:line] — description
- LOW: [file:line] — description

### Unused Dependencies
- [package] — not referenced in any source file

---

Known Stubs (as of 2026-02-26)

These were identified in the last audit and are being tracked:

File	Line	Issue	Severity	Status
server/routes/whatsapp.ts	~20	TODO: orgId from phone number	HIGH	open
server/routes/whatsapp.ts	~24	Hardcoded Spanish response, never sends	CRITICAL	open
server/routes/lemonStub.ts	~8	Entire route is a stub, always returns mock	HIGH	open
server/routes/darya.ts	~180	Feedback not persisted to DB	MEDIUM	open
src/pages/CommandInput.tsx	~45	setTimeout mock instead of real job queue	HIGH	open
src/pages/Files.tsx	all	No data fetching, all empty states, Upload disabled	HIGH	open
server/routes/devika.ts	~8	Phase gates in-memory only (reset on restart)	HIGH	open
src/pages/Content.tsx	~30	Hardcoded orgId: "default-org"	MEDIUM	open
src/pages/Analytics.tsx	~5	Stats from mockMetrics.ts (hardcoded numbers)	MEDIUM	open