---
id: context.analyze-command-execution
version: 1.0.0
category: context
experimental: true
name: Analyze Command Execution
description: Extract detailed execution metadata from each command in the workflow chain
tags:
  - workflow-analysis
  - feedback
  - execution-metrics
model_requirements:
  min_context: 128000
  recommended:
    - claude-sonnet-4.6
    - claude-sonnet-5
agents:
dependencies:
  requires:
    - context.identify-completed-workflow
    - context.use-modern-cli-tools
inputs:
  - name: workflow_executed
    description: Workflow type from previous stage
    type: string
    required: true
  - name: commands_chain
    description: Chain of commands from previous stage
    type: array
    required: true
outputs:
  - agents_used
  - prompts_executed
  - tools_invoked
  - files_changed
  - errors_encountered
  - retries_performed
tokens:
  avg: 3000
  max: 6000
  min: 1500
---

# Analyze Command Execution

## Objective

Extract detailed execution metadata from each command including agents, prompts, tools, file changes, errors, and retries.

## Instructions

### Available Tools

**Use these tools to gather information:**

1. **query_session** - Access session data safely:
   - `action: "get", session_id: "<id>"` - Get command execution details
   - `action: "search", query: "<command-name>"` - Find command executions

2. **Git commands** via `run_terminal_cmd`:
   - `git log`, `git diff`, `git show` - Analyze changes

3. **read_file** - Read definitions:
   - `data/commands/*.md` - Command definitions
   - `data/agents/*.md` - Agent definitions

**DO NOT use:**

- `read_file` on `.valora/sessions/` - use `query_session` instead
- `npm run lint`, `npm test` - not needed for analysis

### Step 1: Identify Agents and Prompts

For each command in `commands_chain`:

**Map command to agent:**

- Use agent responsibilities from WORKFLOW.md
- Examples: `plan` → `@lead`, `implement` → `@<area>-engineer`, `test` → `@qa`

**Identify prompts executed:**

- Load command definition from `data/commands/<command>.md`
- Extract prompt pipeline stages
- List all prompts by stage (context._, code._, review._, documentation._)

### Step 2: Analyze Tool Usage

Parse git diff and commit messages to infer tools used.

**Read operations:**

```bash
rg "read|search|codebase_search" <(git log --stat --oneline HEAD~<n>..HEAD)
```

**Write operations:**

```bash
git diff --numstat HEAD~<n>..HEAD | awk '{print $1, $2, $3}'
```

**Structured commit analysis:**

```bash
# Machine-readable commit data for programmatic processing
git log --format='%H|%an|%s' HEAD~<n>..HEAD
```

Count by tool category:

- `read_file`: Files examined
- `write` / `search_replace`: Files created/modified
- `codebase_search` / `grep`: Search operations
- `run_terminal_cmd`: Commands executed
- `mcp_github_*`: GitHub operations

### Step 3: Track File Changes

```bash
# Structured output for categorization
git diff --name-status HEAD~<n>..HEAD
```

Categorize files:

- **Created** (A): New files
- **Modified** (M): Changed files
- **Deleted** (D): Removed files
- **Renamed** (R): Renamed files

Group by directory to identify affected areas.

### Step 4: Detect Errors and Retries

**From commit messages (use `rg` for pattern matching):**

```bash
# Detect error-resolution patterns in commit messages
git log --format='%s' HEAD~<n>..HEAD | rg -i "fix|resolve|correct|retry|revert"
```

- Look for "fix", "resolve", "correct" keywords
- Identify fixup/amend commits
- Count retry patterns

**From git history:**

- Multiple commits in short succession on same files
- Revert commits
- Emergency fixes

Categorize errors:

- `linter_error`: Linting issues
- `type_error`: Type checking failures
- `test_failure`: Test failures
- `build_error`: Build/compilation errors
- `runtime_error`: Runtime issues

### Step 5: Calculate Retry Metrics

**Retry indicators:**

- Fixup commits
- Same file modified multiple times in succession
- Error keywords in commit messages
- Time gaps suggesting iteration

Count total retries and success rate.

## Output Format

```json
{
	"agents_used": ["product-manager", "lead", "backend-engineer-api", "qa"],
	"prompts_executed": [
		"context.analyze-task-context",
		"plan.breakdown-implementation",
		"code.implement-changes",
		"review.validate-completeness"
	],
	"tools_invoked": {
		"read_file": 45,
		"write": 8,
		"search_replace": 12,
		"codebase_search": 5,
		"run_terminal_cmd": 3,
		"mcp_github_*": 2
	},
	"files_changed": {
		"created": ["src/auth/oauth.ts", "tests/auth/oauth.test.ts"],
		"modified": ["src/auth/index.ts", "package.json"],
		"deleted": [],
		"renamed": []
	},
	"affected_areas": {
		"src/auth": 3,
		"tests/auth": 1,
		"root": 1
	},
	"errors_encountered": [
		{
			"command": "implement",
			"stage": "code",
			"error_type": "linter_error",
			"message": "Unused variable 'token'",
			"file": "src/auth/oauth.ts",
			"resolved": true
		}
	],
	"retries_performed": 2,
	"retry_success_rate": 1.0
}
```

## Success Criteria

- ✅ All agents identified from command chain
- ✅ Prompts extracted from command definitions
- ✅ Tool usage quantified accurately
- ✅ File changes categorized completely
- ✅ Errors detected and classified
- ✅ Retry count calculated

## Error Handling

- **Command definition missing**: Use default agent mapping
- **Tool usage unclear**: Mark as "estimated"
- **No errors detected**: Return empty array (valid state)
- **Cannot parse commits**: Use git log only