Run any Skill in Manus with one click

iblai-api-agent-sandbox

Stars15

Forks3

UpdatedJuly 21, 2026 at 15:29

Connect/disconnect an ibl.ai agent's sandbox (Claw) instance, manage Claw instances, push config, run health/connectivity checks, and set the agent's model — via the platform API. Use when wiring an agent to a sandboxed workspace.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

iblai

iblai/api

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

SKILL.md

readonly

More from this repository

same repository

iblai-api-agent-embed

iblai/api

Configure an ibl.ai agent's embed/widget settings via the platform API — anonymous access, custom CSS/JS, mode, voice/attachment toggles, starter prompts, SSO, share links, and redirect tokens. Use when embedding an agent on an external website.

2026-07-2115

iblai-api-agent-eval

iblai/api

Measure and improve agent quality via the platform API — evaluation datasets, dataset items (JSON, CSV upload, or from chat traces), experiment runs, LLM-as-Judge and human-annotation scoring, score configs, and CSV export. Use to test an agent against a dataset and grade the results.

2026-07-2115

iblai-api-agent-mcp

iblai/api

MCP connectors for ibl.ai agents, end to end — register MCP servers (featured multi-tenant sharing, enable/disable, OAuth service linkage), manage credential connections with the auth_type × scope matrix (org / agent / per-user), wire servers onto an agent, run the OAuth connected-service lifecycle, handle per-user in-chat OAuth (the oauth_required / oauth_connection_resolved chat events), and debug MCP auth failures (401s, missing OAuth prompts, agents ignoring servers). Use when wiring an agent to external MCP servers and tools, or designing/troubleshooting MCP auth.

2026-07-2115

iblai-api-agent-memory

iblai/api

Manage an ibl.ai agent's memories via the platform API — list and filter agent (mentor) memories (by category, user, email, date), curate global (cross-agent) memories, add/edit/delete memories, manage memory categories, and toggle capture/recall settings. Use when inspecting or curating what an agent remembers.

2026-07-2115

iblai-api-crm

iblai/api

Manage an ibl.ai organization's CRM via the platform API — people, organizations, pipelines and stages, lead sources, deals (with stage-move/won/lost actions), activities, and tags. Org-wide sales/relationship management. Use for lead capture, pipelines, and deal flow.

2026-07-2115

iblai-api-notification

iblai/api

Read and send ibl.ai platform notifications via the API — unread counts, the notifications list with filters, mark-as-read (all, bulk, or specific), delete, the two-step notification builder (preview then send/schedule, with source validation and recipient preview), per-user notification preferences, per-user push-device (FCM) tokens, and org-admin email-template & SMTP management. Use when reading an organization's notifications, sending one, tuning a user's preferences, or configuring notification templates.

2026-07-2115

name	iblai-api-agent-sandbox
description	Connect/disconnect an ibl.ai agent's sandbox (Claw) instance, manage Claw instances, push config, run health/connectivity checks, and set the agent's model — via the platform API. Use when wiring an agent to a sandboxed workspace.

iblai-api-agent-sandbox

Wire an agent to a Claw sandbox instance, manage the available instances, push config and run health/connectivity checks, and set the model the agent's workspace uses. Use when connecting an agent to a sandboxed workspace.

Auth & conventions

Base URL: https://api.iblai.app
Header: Authorization: Api-Token $IBLAI_API_KEY on every request.
Prefix: paths shown as … are under https://api.iblai.app/dm/api/ai-mentor — the /dm gateway routes to the Data Manager service. (…/api/ai-agent/… is an accept-only alias for the same routes.)
Path vars: {org} = $IBLAI_ORG, {username} = $IBLAI_USERNAME, {mentor} = the agent's unique id (a UUID, e.g. d17dc729-60fd-4363-81a0-f67d9318b03e). {id} on claw/instances/ and claw/model-providers/ is the integer primary key.
Not connected yet? Run /iblai-api-login first to populate IBLAI_ORG, IBLAI_USERNAME, and IBLAI_API_KEY.

Reads

Instances

GET …/orgs/{org}/claw/instances/ — list registered Claw instances.
GET …/orgs/{org}/claw/instances/{id}/ — retrieve one instance (status, claw_version, last health/pairing).

Model providers

GET …/orgs/{org}/claw/model-providers/ — list the org's model-provider blocks.
GET …/orgs/{org}/claw/model-providers/{id}/ — retrieve one provider block.

Agent config

GET …/orgs/{org}/mentors/{mentor}/claw-config/ — the agent's connected instance + push status (404 = not connected).
GET …/orgs/{org}/mentors/{mentor}/agent-config/ — agent workspace config (includes current model).
GET …/orgs/{org}/users/{username}/mentor-llms/?mentor_id={mentor} — available model options.

Writes

Instances

POST …/orgs/{org}/claw/instances/ — register an instance:

{
  "name": "string",
  "claw_type": "openclaw | ironclaw | nemoclaw",
  "server_url": "https://claw.example.com",
  "gateway_token": "string (write-only)",
  "connection_params": { "device_identity": {} },
  "auth_headers": {},
  "deployment_backend": "backend id (optional)"
}

PATCH …/claw/instances/{id}/ — edit (send only changed writable keys: name, claw_type, server_url, gateway_token, connection_params, auth_headers, deployment_backend).
DELETE …/claw/instances/{id}/ — delete instance. Destructive — confirm with the user first.
POST …/claw/instances/{id}/health-check/ — queue a health check (returns 202; poll the instance for last_health_status).
POST …/claw/instances/{id}/test-connectivity/ — TLS reachability + health check.
POST …/claw/instances/{id}/refresh-version/ — detect claw_version from the instance handshake.
POST …/claw/instances/{id}/request-pairing/ — queue a device-pairing request (pending request expires ~5 min). Outward-facing — confirm with the user first.
POST …/claw/instances/{id}/push-providers/ — push all enabled model providers to the instance. Outward-facing — confirm with the user first.
POST …/claw/instances/{id}/security-audit/ — run a security audit (OpenClaw only). Outward-facing — confirm with the user first.

Model providers

POST …/orgs/{org}/claw/model-providers/ — register a provider block:

{
  "server": "instance id (required)",
  "name": "minimax",
  "base_url": "https://api.minimax.io/anthropic",
  "api_type": "anthropic-messages",
  "credential_name": "string",
  "credential_key": "key",
  "model_catalog": [{ "id": "...", "name": "..." }],
  "models_mode": "merge | replace",
  "enabled": true
}

PATCH …/claw/model-providers/{id}/ — edit a provider (send only changed keys).
DELETE …/claw/model-providers/{id}/ — delete a provider. Destructive — confirm with the user first.

Agent config

POST …/mentors/{mentor}/claw-config/ — connect the agent to an instance:

{
  "server": "instance id (required)",
  "enabled": true,
  "auto_push": false,
  "agent_config": {}
}

PATCH …/mentors/{mentor}/claw-config/ — edit the connection (enabled, auto_push, agent_config).
DELETE …/mentors/{mentor}/claw-config/ — disconnect. Destructive — confirm with the user first.
POST …/mentors/{mentor}/claw-config/push-config/ — push the agent's agent_config to the instance. Outward-facing — confirm with the user first.
PATCH …/mentors/{mentor}/agent-config/ — update the agent's workspace config (send only changed keys):
```
{
  "model": "string",
  "allowed_models": "string[]",
  "identity": "string", "soul": "string", "user_context": "string",
  "tools": "string", "agents": "string", "bootstrap": "string",
  "heartbeat": "string", "memory": "string",
  "config": {}
}
```
identity / soul / user_context / tools / agents / bootstrap / heartbeat / memory are the claw workspace files (IDENTITY.md, SOUL.md, …) pushed to the instance; config is a JSON instance-config patch (heartbeat schedule, session isolation, skill toggles, model fallbacks). config is deny-list validated — writes to gateway.auth, gateway.controlUi.dangerouslyDisableDeviceAuth, tools.exec.host, sandbox.mode, and hooks.allowUnsafeExternalContent are rejected.

Example

Connect an agent to a Claw instance and enable it (server is the instance's integer id):

curl -X POST \
  "https://api.iblai.app/dm/api/ai-mentor/orgs/$IBLAI_ORG/mentors/$MENTOR/claw-config/" \
  -H "Authorization: Api-Token $IBLAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"server": 12, "enabled": true}'

Notes

Ids differ by resource: {mentor} is a UUID, but server and {id} on claw/instances/ and claw/model-providers/ are integer primary keys.
A 404 from claw-config/ means the agent is not connected to a sandbox yet — use the connect POST to wire it up.
OpenClaw device identity: include an Ed25519 keypair under connection_params.device_identity when registering an OpenClaw instance, or config push fails with missing scope: operator.read.
health-check is queued (202) — poll the instance for last_health_status. Run test-connectivity before connecting an agent to confirm the sandbox is reachable.
request-pairing mints a pending request that must be approved on the instance side (once per platform connection); security-audit is OpenClaw-only.
After connecting, push-config (or set auto_push on the claw-config) so the agent's workspace reflects the latest agent_config; push-providers syncs the org's enabled provider blocks to the instance.
claw_type is openclaw, ironclaw, or nemoclaw; gateway_token authenticates the instance to its server. gateway_token, auth_headers, and connection_params are write-only (never returned by reads).