Prompt file imported from Coin333/claude-context (.claude/commands/trace.md). Fill in {{arguments}} before use. Copyright stays with the author.

Trace

Trace the flow of {{arguments}} through the codebase. Map every file and function involved, identify where data transforms or could break, and report the full chain.

Step 1: Identify the flow type

Determine which category applies:

• HTTP request — from route definition → middleware → handler → DB → response
• Data mutation — from UI action → API call → validation → DB write → side effects
• Background job/agent — from trigger → queue → worker → output/notification
• Auth flow — from token → middleware → user resolution → authorization check
• Event/webhook — from inbound payload → validation → processing → response

Step 2: Find the entry point

Search for the entry point related to {{arguments}}:

grep -rn "{keyword}" src/ --include="*.ts" --include="*.js" --include="*.py" -l

Read the most likely entry point file in full.

Step 3: Trace the chain

Follow the code path step by step:

1. Read each function in the chain
2. Note where data changes shape (transforms, parses, validates)
3. Note failure points — where could data be lost, malformed, or misrouted?
4. Identify ownership boundaries — where does one module hand off to another?

Step 4: Report the chain

Entry: {file:line} — {function name}
  ↓ {transform or action}
  {file:line} — {function name}
  ↓ {transform or action}
  ...
Exit: {file:line} — {where the data ends up / what is returned}

Then list:

• Data transforms: how the data changes shape at each step
• Failure points: where it could break and what the failure mode would be
• Findings: any bugs, missing validation, or suspicious patterns
• Recommendation: what to fix or investigate further

Be specific about file paths and line numbers. Never report a finding without verifying the actual code.