Implement a feature following TDD workflow. $ARGUMENTS is a spec file path (e.g., specs/my-feature.md) or a plain-text feature description.

Scope: This skill writes Go code and tests. It does NOT update website docs (use update-docs after) or CHANGELOG (use changelog after).

Workflow

Step 1: Understand Requirements

If $ARGUMENTS is a file path:

1. Read the spec file
2. Extract acceptance criteria and edge cases
3. Identify affected packages

If $ARGUMENTS is a description:

1. Search existing code for related functionality
2. Identify the right package to extend
3. Confirm scope with user before proceeding

Step 2: Identify Affected Files

List all files that will be created or modified:

# Typical pattern for a new command
cmd/skillshare/<command>.go          # Command handler
cmd/skillshare/<command>_project.go  # Project-mode handler (if dual-mode)
internal/<package>/<feature>.go      # Core logic
tests/integration/<command>_test.go  # Integration test

Display the file list and continue. If scope is unclear, ask the user.

Step 3: Write Failing Tests First (RED)

Write integration tests using testutil.Sandbox:

func TestFeature_BasicCase(t *testing.T) {
    sb := testutil.NewSandbox(t)
    defer sb.Cleanup()

    // Setup
    sb.CreateSkill("test-skill", map[string]string{
        "SKILL.md": "---\nname: test-skill\n---\n# Content",
    })

    // Act
    result := sb.RunCLI("command", "args...")

    // Assert
    result.AssertSuccess()
    result.AssertOutputContains("expected output")
}

Verify tests fail:

make test-int
# or run specific test:
go test ./tests/integration -run TestFeature_BasicCase

Step 4: Implement (GREEN)

Write minimal code to make tests pass:

1. Follow existing patterns in cmd/skillshare/ and internal/
2. Use internal/ui for terminal output (colors, spinners, boxes)
3. Add oplog instrumentation for mutating commands:

   start := time.Now()
   // ... do work ...
   e := oplog.NewEntry("command-name", statusFromErr(err), time.Since(start))
   oplog.Write(configPath, oplog.OpsFile, e)

4. Register command in main.go commands map if new command

Verify tests pass:

make test-int

Step 5: Refactor and Verify

1. Clean up code while keeping tests green
2. Run full quality check:

   make check  # fmt-check + lint + test

3. Fix any formatting or lint issues

Project Patterns Reference

These patterns appear throughout the codebase. Follow them when implementing new features.

Handler Split Convention

Large commands are split by concern rather than kept in a single file. When a command handler grows beyond ~300 lines, split it:

Principle: dispatch file does ONLY flag parsing + mode routing. Logic goes in sub-files.

Dual-Mode Command Pattern

Most commands support both global (-g) and project (-p) mode:

func handleMyCommand(args []string) error {
    mode, rest, err := parseModeArgs(args)
    if err != nil { return err }

    switch mode {
    case modeProject:
        return handleMyCommandProject(rest)
    default:
        return handleMyCommandGlobal(rest)
    }
}

Create <cmd>_project.go for project-mode handler. Use parseModeArgs() from mode.go.

TUI Components (bubbletea)

All interactive prompts use bubbletea (not survey). Key components:

• checklist_tui.go — shared checklist/radio picker
• list_tui.go — filterable list with detail panel
• search_tui.go — multi-select checkbox list

Color palette: cyan Color("6"), gray Color("8"), yellow #D4D93C.

Dispatch order: JSON output → TUI (if TTY + items + !--no-tui) → empty check → plain text.

Web API Endpoint

If the feature needs a Web UI endpoint, add internal/server/handler_<name>.go:

func (s *Server) handle<Name>(w http.ResponseWriter, r *http.Request) {
    // ...
    writeJSON(w, result)       // 200 OK with JSON
    // writeError(w, 400, msg) // for errors
}

Register in server.go route setup. Branch on s.IsProjectMode() for mode-specific behavior.

Oplog Instrumentation

All mutating commands log to operations.log (JSONL):

start := time.Now()
// ... do work ...
e := oplog.NewEntry("command-name", statusFromErr(err), time.Since(start))
e.Args = map[string]any{"key": value}
oplog.Write(configPath, oplog.OpsFile, e)

Security scans write to oplog.AuditFile instead.

Step 6: E2E Runbook (Major Features Only)

If the feature meets any of these criteria, generate an E2E runbook:

• New command or subcommand
• Changes to install/uninstall/sync flow
• Security-related (audit, hash verification, rollback)
• Multi-step user workflow (init → install → sync → verify)
• Edge cases that integration tests alone can't cover (Docker, network, file permissions)

Generate ai_docs/tests/<slug>_runbook.md following the existing convention:

# CLI E2E Runbook: <Title>

<One-line summary of what this validates.>

**Origin**: <version> — <why this runbook exists>

## Scope

- <bullet list of behaviors being validated>

## Environment

Run inside devcontainer with `ssenv` isolation.

## Steps

### 1. Setup: <description>

\```bash
<commands>
\```

**Expected**: <what should happen>

### 2. <Action>: <description>
...

## Pass Criteria

- All steps marked PASS
- <additional criteria>

Key conventions:

• YAML-free, pure Markdown
• Each step has bash block + Expected block
• ss = skillshare, ~ = ssenv-isolated HOME
• Runbook can be executed by the cli-e2e-test skill

If the feature does not meet the criteria above, skip this step.

Step 7: Stage and Report

1. List all created/modified files
2. Confirm each acceptance criterion is met with test evidence
3. Remind user to run update-docs if the feature affects CLI flags or user-visible behavior

Rules

• Test-first — always write failing test before implementation
• Minimal code — only write what's needed to pass tests
• Follow patterns — match existing code style in each package
• 3-strike rule — if a test fails 3 times after fixes, stop and report what's blocking
• No docs — this skill writes code only; use update-docs for documentation
• No changelog — use changelog skill for release notes
• Spec ambiguity — ask the user rather than guessing