veloran
v0.2.21
Published
Scaffold and validate multi-vendor coding-agent harness workflows
Readme
Veloran
Veloran installs a coding-agent harness for Codex, Claude Code, OpenCode, Google Antigravity, and AGENTS.md-compatible tools.
It gives agents the files they need to prepare a real local workflow: instructions, skills, ExecPlans, context, memory, knowledge, harness scripts, observability scaffolding, and validation docs.
Install
Run this from the repository you want to prepare:
npx -y veloran@latestThe installer shows menus for install scope and app targets. For most users, choose the recommended Global user install and the coding apps you use.
Use The Skills
After install, open your coding agent and mention the installed skill:
@init-harnessto inspect the repo and prepare startup, logs, metrics, traces, browser checks, config examples, memory, knowledge, and validation evidence.@execplan-createto create a self-contained implementation plan.@execplan-executeto execute a saved plan milestone by milestone.
If your app does not autocomplete skills, ask it to use the skill by name, such as use the init-harness skill.
What Veloran Adds
- Root guidance files for the selected apps.
- Skills for harness initialization and ExecPlan workflows.
.agent/knowledge/for indexed rules, standards, facts, and docs..agent/memory/and.agent/context/to avoid repeated discovery..agent/harness/scaffolding for local startup, observability, and validation.
Knowledge And Safety
Veloran knowledge composes from user-global knowledge to repository knowledge to path-specific knowledge. Agents should default new facts and standards to the local repository, and only write global knowledge with user approval or repeated cross-repository evidence.
Do not store secrets, tokens, passwords, private keys, customer data, raw production logs, or personal data in knowledge, memory, context, plans, prompts, docs, transcripts, or validation logs.
