Vibeos Api Automation

Call VibeOS API server from scripts via vibeos api.

Skill metadata

Source	Bundled (installed by default)
Path	skills/devops/vibeos-api-automation
Version	1.0.0
Author	VibeOS
License	MIT
Platforms	linux, macos, windows
Tags	api, sdk, automation, ci, runs, openai
Related skills	cursor-capabilities-bridge, vibeos-local-verify-deploy, cursor-sdk-bridge, analytical-ui

Reference: full SKILL.md

info

The following is the complete skill definition that VibeOS loads when this skill is triggered. This is what the agent sees as instructions when the skill is active.

VibeOS API Automation

Use the existing OpenAI-compatible API server through the thin vibeos api CLI — Cursor-SDK-like invoke without new core tools.

When to Use

• Scripts / CI need to start an agent run and poll/stream results
• User asks for “programmatic agent”, “SDK”, “HTTP API”, “automate vibeos”
• Connecting Open WebUI / curl / custom bots to a running gateway

Prerequisites

In ~/.vibeos/.env (or coding profile .env):

API_SERVER_ENABLED=true
API_SERVER_KEY=change-me-local-dev
# optional: API_SERVER_PORT=8642

Gateway must be running: vibeos gateway (API listens on :8642 by default).

How to Run

vibeos api health
vibeos api chat "Summarize AGENTS.md in 3 bullets"
vibeos api run "Run targeted tests for vibeos_cli/api_cmd.py" --wait
RUN=$(vibeos api run "…" | jq -r .run_id)
vibeos api stream "$RUN"
vibeos api stop "$RUN"
SID=$(vibeos api session-create --title demo | jq -r .id // .session_id)
vibeos api prompt --session "$SID" "continue the previous task"

Overrides: --base-url, --api-key, or env VIBEOS_API_BASE / API_SERVER_KEY.

Quick Reference

Goal	Command
Health	vibeos api health[--detailed]
Sync chat	vibeos api chat "…"
Async run	vibeos api run "…" [--wait|--follow]
Poll run	vibeos api get <run_id>
SSE	vibeos api stream <run_id>
Stop	vibeos api stop <run_id>
Sessions	vibeos api sessions / session-create
Resume	vibeos api prompt --session <id> "…"

Procedure

1. Confirm API server enabled and gateway up (vibeos api health).
2. Prefer run + stream/get for long agent work; chat for short Q&A.
3. For multi-turn bots, create a session and pass --session / prompt.
4. Never commit API_SERVER_KEY; keep it in .env only.
5. Do not add new core tools for HTTP invoke — this CLI is the footprint-safe path.

Pitfalls

• API server refuses to start without API_SERVER_KEY
• Background delegate_task / terminal notify may not deliver back over HTTP (supports_async_delivery=False on api_server) — use runs/SSE instead
• GitHub push remains blocked by vibeos-policy even when called via API

Verification

• vibeos api health returns HTTP 200 JSON
• vibeos api chat "ping" returns assistant text
• vibeos api run "ping" --wait reaches completed (or clearly fails with JSON error)