Safety-first agentic coding framework. Three-layer safety pipeline (sandbox, YAML policies, human-in-the-loop approval) for AI coding agents. Pluggable runtimes (Claude Code, Codex), autonomous task orchestrator, full audit trail.
Latest release v2.5.1 · 23 Oct 2026
9 files for Claude Code: architecture, debug-leashd, debug-task, telegram-harness and 5 more — 1,144 tokens loaded in every session.
CLAUDE.md A 1,144 tok .claude/settings.json A — .claude/skills/architecture/SKILL.md C 74 tok .claude/skills/debug-leashd/SKILL.md A 49 tok .claude/skills/debug-task/SKILL.md A 51 tok .claude/skills/telegram-harness/SKILL.md A 90 tok .claude/skills/tmux-runtime/SKILL.md B 116 tok .claude/skills/heal-e2e/SKILL.md A 33 tok .claude/skills/release-prep/SKILL.md A 127 tok These files are vmehera123/leashd's own configuration — they tell Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.