code-trace

Codex Compatibility

When running this skill in Codex, translate Claude Code-only primitives before acting: AskUserQuestion -> chat/requestuserinput, TodoWrite -> updateplan, Task/TaskCreate/TeamCreate/SendMessage -> spawnagent/sendinput/waitagent when available and allowed, and EnterPlanMode/ExitPlanMode -> a concise chat plan plus explicit approval. Resolve Read/Write/Edit/Bash/WebSearch/WebFetch to Codex file/shell/web tools, and map ~/.claude/... paths to ~/.agents/... or ~/.codex/... unless the task explicitly targets Claude Code.

Cursor Compatibility

When running this skill in Cursor Agent, translate Claude Code-only primitives before acting: AskUserQuestion -> AskQuestion; TodoWrite -> Cursor TodoWrite or an equivalent checklist; Task/TaskCreate/TeamCreate/SendMessage/multi-agent flows -> Cursor Task (subagents), parallel Tasks, or runinbackground when allowed (TeamCreate/SendMessage may have no exact match); EnterPlanMode/ExitPlanMode -> Plan mode (SwitchMode / CreatePlan) plus explicit user approval. Resolve Read/Write/Edit/StrReplace/Bash/web/search/MCP via Cursor Composer or Agent equivalents. MCP names written as mcpservertool typically map to callmcptool with configured server identifiers. Map ~/.claude/... to ~/.cursor/skills/, .cursor/skills/, and .cursor/rules/ unless the task explicitly targets Claude Code.

<essential_principles>

How Code Tracing Works

This skill traces code execution paths interactively, letting you navigate through the codebase like a debugger stepping through code - but with rich explanations at each step.

Principle 1: Application Boundary

Trace ONLY application code. External dependencies (node_modules, vendor/) receive:

• A summary of what they do
• Link to official documentation
• NOT deep-traced into their internals

Why: External libraries can be 100K+ lines. Tracing into them wastes context and obscures the actual application logic. The goal is understanding YOUR code, not library internals.

Principle 2: Interactive Navigation

Every conditional branch becomes a user choice:

Why: Linear traces miss important paths. Interactive navigation lets users explore exactly what they're interested in.

Principle 3: Progressive Explanation

Each step includes:

1. Location: File + line range + function name
2. Code: Full source (no abbreviation)
3. What: Brief summary of what the code does
4. Why: Why this step exists in the flow
5. Next: What happens next (or choices if branching)

Use thinking markers (🤔🎯⚡📊💡🔐) for clarity.

Principle 4: State Persistence

Trace state is stored in Serena Memory to enable:

• Resuming interrupted traces
• Backtracking to previous decision points
• Saving completed traces for future reference

</essential_principles>

<intake>

What would you like to trace?

1. Trace a request flow - Follow HTTP request from receipt to response
2. Trace a function call - Follow a specific function through the codebase
3. Resume previous trace - Continue from where you left off

Please provide additional context:

• For request tracing: Which endpoint? (e.g., "POST /api/users")
• For function tracing: Which function? (e.g., "validateUser" or "src/utils/auth.ts:checkToken")

Wait for response before proceeding. </intake>

<routing>

Before Starting Any Workflow

1. Detect Framework: Run scripts/detect-framework.sh to identify:

- Express, Next.js (App/Pages), Fastify, Hono, NestJS, Koa, or generic

1. Load Framework Patterns: Read references/framework-patterns.md section for detected framework

1. Prepare Serena: Ensure Serena MCP is available for:

- findsymbol() - Locate functions/handlers - findreferencingsymbols() - Find callers - getsymbolsoverview() - Map module structure - writememory() / read_memory() - State persistence

After determining intent and framework, read the appropriate workflow and follow it. </routing>

<reference_index>

References

All in references/:

</reference_index>

<workflows_index>

Workflows

All in workflows/:

</workflows_index>

<scripts_index>

Scripts

Usage:

./scripts/detect-framework.sh /path/to/project
# Output: express | nextjs-app | nextjs-pages | fastify | hono | nestjs | koa | generic

</scripts_index>

<success_criteria> A successful code trace:

• Entry point correctly identified and explained
• Framework detected and appropriate patterns applied
• At least one branch point presented as interactive choice
• External dependencies summarized (not deep-traced)
• User navigated to terminal point OR chose to stop
• Path history shown in ASCII flowchart format
• Mermaid flowchart offered as output option (if trace completed)
• Key insights collected and displayed
• Trace state available for resume (if user chose to save)

</success_criteria>

<boundaries>

Boundaries

Will:

• Trace application code with full source display
• Explain each step with thinking markers
• Present conditional branches as interactive choices
• Summarize external dependencies at the boundary
• Persist trace state for resume capability

Will Not:

• Deep-trace into node_modules or external libraries
• Execute or run the code (read-only analysis)
• Modify any source files
• Make assumptions about runtime values (present all branches)

</boundaries>