dev-autopilot and pilot
An interactive planning orchestrator (A) and a fully autonomous ticket-to-deployment pipeline (B) serve complementary workflow stages—one prioritizes human oversight during planning while the other emphasizes end-to-end autonomy—making them alternatives rather than companions for the same development process.
About dev-autopilot
AsiaOstrich/dev-autopilot
Agent-agnostic unattended development orchestrator. Plan interactively, execute autonomously, review results.
# Technical Summary Implements a DAG orchestrator that converts structured task specifications into parallel/sequential execution pipelines across pluggable AI agents (Claude, OpenCode, CLI) with built-in safety hooks, quality gates, and a fix-loop that auto-retries failed tasks within budget constraints. The architecture uses a three-phase workflow—spec generation via `/sdd`, plan compilation via `/plan`, and autonomous execution via `/orchestrate`—with optional AI judge review and topological task dependency resolution for optimized parallelization.
About pilot
alekspetrov/pilot
AI that ships your tickets - autonomous dev pipeline with Claude Code
claude-code (Anthropic) 🏃 Active: ping-app (6m 23s) tokens/cost 📊 Today: 14 tasks, 284.2M tokens, $3.42 ┌──────────────────────────────────────────────┐ │ Task Queue │ ├──────────────────────────────────────────────┤ │ 🟢 Implementing "Add rate limiting" (6m 23s) │ │ 🟡 Queued (4) - Next in 3m │ │ 🔴 Failed (2) │ └──────────────────────────────────────────────┘ ┌──────────────────────────────────────────────┐ │ Autopilot: stage (CI
Scores updated daily from GitHub, PyPI, and npm data. How scores work