Auto-Claude

Auto-Claude

by AndyMik90

Autonomous multi-session AI coding companion — plan, code, and validate features with autonomous agents.

See https://github.com/AndyMik90/Auto-Claude

Features

• Autonomous agents that plan tasks, generate code, and run validations (spec-driven + self-review).
• Multi-session / multi-agent architecture (parallel agent terminals) so multiple tasks can be worked on concurrently.
• Kanban-style task board and task management UI for organizing work and tracking progress.
• Context & memory system: programmatic project analysis to maintain persistent context across sessions.
• GitHub integration: pull issues, create branches/PRs, and generate changelogs from completed tasks.
• Changelog generation and traceability: audit trail between tasks, commits, and issues.
• Roadmapping and feedback integration (e.g., Canny) to prioritize work according to user feedback.

Superpowers

Auto-Claude is aimed at developers and teams who want to accelerate delivery while keeping full visibility and control. Its strengths are:

• Spec-driven autonomous work: agents not only write code but also validate and self-review before producing changes.
• Parallelism: run several agents concurrently on independent tasks (feature work, bug fixes, tests) to reduce turnaround time.
• Persistent project understanding: the memory/context system prevents repeated re-explaining of the codebase, enabling longer-lived agent sessions.
• GitOps-friendly outputs: agents produce commits, PRs, and changelogs that integrate with GitHub workflows for easy review and CI testing.

Who this is for

• Individual devs who want an AI pair programmer that can take on whole tasks (not just single-line completions).
• Small teams wanting to prototype features quickly and iterate with automated spec checks.
• Projects that benefit from auditability — traceable agent decisions, changelogs and PRs.

Practical usage examples (no install steps)

• Bug triage & fix: point Auto-Claude at a GitHub repo or sync issues. It can create a task from an issue, spin up an agent to reproduce and fix the bug, run unit tests, and open a PR with a descriptive changelog and test evidence.

• Feature implementation workflow: add a feature card on the Kanban board. Auto-Claude will break the feature into sub-tasks, run agents to implement components in parallel (frontend, backend, tests), validate them, and assemble a single PR with passing tests and an automated spec summary.

• Continuous backlog processing: connect a feedback feed (Canny or similar). Auto-Claude will analyze feedback, map requests to product goals, and produce prioritized tasks and a roadmap view — with agents proposing minimal viable implementations.

Integration notes

• GitHub: supports issue sync and PR creation. Useful in GitOps pipelines — agents create branches and PRs which are then validated by your CI.
• Feedback/roadmap tools: integrates ideas/feedback into backlog items for agent-driven implementation and prioritization.

Known limitations & issues

• Reported GitHub integration problems in some environments (macOS auth flow, and issues accessing organization repositories) — check the repo issues for current status.
• Quality depends on available context and tests: projects with sparse tests or poor docstrings will require more human oversight.
• Agent-driven changes still need security and dependency review before merging in production systems.
• As an autonomous system, it can produce incorrect or suboptimal designs; keep review gates and CI checks enabled.

Pricing

• Open-source / free to use (see repository for project license and contribution guidelines). Commercial hosting or managed variants may exist via forks or 3rd-party services.

Community & development

• Source and issues: https://github.com/AndyMik90/Auto-Claude
• Community forks and variations exist; watch issues and forks for real-world usage reports and fixes.

Sources & further reading

• Project repository: https://github.com/AndyMik90/Auto-Claude
• Repository issues & discussions (for known bugs and integration notes)
• Demos / community videos demonstrating multi-agent Kanban workflows (search “Auto Claude” demos on YouTube)