Batch Commit Processor Agent

Purpose

Process multiple Git commits in batch, extracting smart commit commands, aggregating time logs per issue, deduplicating comments, and executing batch updates to Jira. Designed for efficient processing of commit ranges with graceful error handling.

Core Capabilities

1. Commit Range Parsing

Parse Git commit ranges (HEAD~5..HEAD, main..feature-branch, date ranges)
Extract commit messages and handle various range formats
Support author filters and custom date ranges

2. Smart Command Extraction

Extract #time commands with duration parsing (1h, 30m, 2h30m, 1.5h)
Extract #comment commands with message content
Extract #transition commands for workflow state changes
Parse issue keys from commit messages (PROJ-123, TEAM-456)

3. Time Aggregation

Sum all time entries per issue key
Convert various time formats to seconds with validation
Handle fractional hours and compound durations
Format aggregated time for Jira worklog API

4. Comment Deduplication

Remove exact duplicate comments
Merge similar comments using fuzzy matching (>80% similarity)
Preserve unique information while maintaining chronological order

5. Batch Execution

Group updates by issue key
Execute Jira API calls efficiently with partial failure handling
Support dry-run mode for preview
Provide detailed success/error reporting

Input Format

commit_range: "HEAD~5..HEAD"  # Git range
dry_run: false                 # Preview mode
author_filter: null            # Optional
since: "2025-12-01"           # Optional date
until: "2025-12-19"           # Optional date

Processing Workflow

Phase 1: Extraction - Parse commit messages for issue keys and smart commands Phase 2: Aggregation - Group commands by issue key, sum times, collect comments Phase 3: Validation - Verify issues exist, check transitions, validate time values Phase 4: Execution - Add worklogs, comments, apply transitions Phase 5: Reporting - Generate batch summary with results and errors

Time Duration Parsing

Supported formats: 1h, 30m, 2h30m, 1.5h, 90m, 0.5h, 1h 30m

Conversion logic:

Extract hours portion: multiply by 3600
Extract minutes portion: multiply by 60
Sum for total seconds
Format back to readable h/m format

Comment Deduplication

Strategy:

Exact match: Keep first occurrence
Fuzzy match: Merge similar comments (>80% similarity using Levenshtein distance)
Semantic grouping: Combine related comments into single entry

MCP Tools Usage

Tool	Purpose	Input	Error Handling
getJiraIssue	Verify issue exists	issueIdOrKey	Skip issue, log error
addWorklogToJiraIssue	Add aggregated time	issueIdOrKey, timeSpent	Continue batch, retry with backoff
addCommentToJiraIssue	Add deduplicated comments	issueIdOrKey, commentBody	Continue batch
transitionJiraIssue	Apply workflow transition	issueIdOrKey, transition.id	Skip transition, continue

Output Format

Success Response

batch_result:
  commit_range: "HEAD~5..HEAD"
  commits_processed: 5
  issues_affected: 2
  dry_run: false
  execution_time_ms: 3421
  results:
    PROJ-123:
      status: "success"
      total_time_logged: "3h 30m"
      comments_added: 1
      transition: "In Review"
      commits: ["abc123", "def456", "ghi789"]
    PROJ-456:
      status: "success"
      total_time_logged: "30m"
      comments_added: 1
  summary:
    total_time_logged: "4h"
    total_comments: 2
    success_rate: 100
  errors: []

Partial Failure Response

Includes failed issues with error types: issue_not_found, transition_failed, api_timeout

Dry Run Response

Returns preview with would_log_time, would_add_comments, would_transition_to and validation warnings

Error Handling

Validation Errors: issue_not_found, invalid_time, invalid_transition (skip specific operation, continue) Execution Errors: api_timeout, rate_limit, permission_denied, network_error (retry with exponential backoff) Partial Failures: Continue batch processing, include errors in final report

Retry Strategy: Max 3 attempts, exponential backoff (1s-10s delay), skip non-retryable errors

Configuration

Environment Variables:
  BATCH_COMMIT_DRY_RUN: false
  BATCH_COMMIT_MAX_COMMITS: 100
  BATCH_COMMIT_MAX_TIME_HOURS: 8
  BATCH_COMMIT_SIMILARITY_THRESHOLD: 0.8
  BATCH_COMMIT_RETRY_ATTEMPTS: 3

Agent Config:
  model: sonnet
  timeout_seconds: 300
  max_iterations: 10
  features: [time_aggregation, comment_deduplication, transition_support, dry_run_mode, parallel_validation]

Integration Points

Called By: qa-comment-responder, qa-ticket-reviewer, git hooks, CI/CD pipelines, manual workflow Calls: Git (commit parsing), Jira MCP tools (issue updates), utility functions (time parsing, dedup)

Data Flow: Git commits → Extracted commands → Aggregated data → Validated data → Jira updates → Batch result

Performance Constraints

Recommended batch size: 10-50 commits (max 100)
Max issues per batch: 50
Max time per issue: 8h
Max comment length: 32,767 chars (Jira limit)
Parallel validation for issue checks, sequential execution for updates

Status

Status: Ready for implementation Owner: Jira Orchestration Team Last Reviewed: 2025-12-19

qa-comment-responder.md
qa-ticket-reviewer.md
qa-confluence-documenter.md

batch-commit-processor