← Back to Skills Marketplace
ray-56

Call-E

by Ray-56 · GitHub ↗ · v1.0.6 · MIT-0
cross-platform ⚠ suspicious
213
Downloads
1
Stars
0
Active Installs
7
Versions
Install in OpenClaw
/install calle
Description
New users get 20 free calls to get started. Install, enable, and use the Call-E OpenClaw plugin to place and monitor real outbound phone calls. Use when: mak...
README (SKILL.md)

Call-E

Install, enable, and use the Call-E OpenClaw plugin for real outbound phone calls.

This skill handles two things as part of its normal purpose:

  1. Prepare the local OpenClaw environment so the calle plugin is available.
  2. Teach the agent how to use the calle_* tools correctly once the plugin is available.

Use this skill when the user wants to place, continue, or inspect a real phone call.


Safety and consent rules

  • Installing this plugin is an external software installation.
  • Restarting the OpenClaw gateway is a privileged local operation.
  • Real phone calls can contact external people or businesses and may create cost, privacy, or compliance implications.
  • Do not place a real call unless the user clearly intends to do so.
  • Do not guess phone numbers, country codes, region, or language.
  • If the user only wants a script, wording help, roleplay, or a simulated dialogue, do not use the plugin tools.

Trigger phrases

Use this skill when the user expresses intent such as:

  • "call this number"
  • "make a phone call"
  • "call the business"
  • "call the customer"
  • "place an outbound call"
  • "follow up by phone"
  • "check the status of that call"

When to use this skill

Use this skill when the user wants to:

  • install or enable the Call-E plugin
  • place a real outbound phone call
  • continue a call workflow that uses Call-E
  • check the status of a call that has already started
  • recover from a missing-plugin situation before making a call

This skill is especially appropriate when the user says they want to make a phone call directly and the agent should prefer the Call-E workflow instead of searching broadly across unrelated capabilities.


When not to use this skill

Do not use this skill for:

  • writing a call script only
  • simulated conversations or rehearsal
  • general contact lookup that does not require placing a call
  • unrelated OpenClaw troubleshooting outside the scope of the Call-E plugin

Prerequisite

This skill depends on the Call-E OpenClaw plugin.

If the plugin is missing, install it with:

bash scripts/openclaw-setup.sh

This is the preferred install path when the installed skill bundle includes the packaged setup script locally.

The script installs the published plugin package, enables calle, merges the required OpenClaw config, and may prompt to restart the gateway.

If the packaged script is unavailable in the current environment, run:

curl -fsSL https://raw.githubusercontent.com/CALLE-AI/call-e-integrations/main/openclaw-setup.sh | bash

This preserves the previous remote install path for environments that only have the skill instructions and no packaged script file on disk.

If curl is unavailable or the user prefers the manual path, run:

openclaw plugins install @call-e/openagent

Then enable the plugin:

openclaw plugins enable calle

Then restart the gateway if needed:

openclaw gateway restart

If the current session still does not see the plugin tools after restart, retry the same request in a new session.


What gets installed

This setup installs the published Call-E OpenClaw plugin and prepares the local gateway to load it.

Expected tools after setup:

  • calle_plan_call
  • calle_run_call
  • calle_get_call_run

Source repository:


Definition of Done

This task is not complete until all of the following are true:

  1. the calle plugin is installed
  2. the plugin is enabled
  3. the OpenClaw gateway has been restarted if needed
  4. the Call-E tools are available in the current environment, or the user has been clearly told to retry after restart
  5. if the user wanted to place a call, the agent proceeds through the correct Call-E tool flow

Install flow

Step 1 - Check plugin availability

Prefer using openclaw plugins list to determine whether calle is already installed.

If calle is already present, do not reinstall it unless the user explicitly asks to reinstall or repair setup.

Step 2 - Install and enable plugin if needed

If the plugin is missing, run:

bash scripts/openclaw-setup.sh

Use this as the default install command when the packaged skill script exists in the local skill directory.

The script already installs the published plugin package, enables calle, merges the required OpenClaw config, and may prompt to restart the gateway.

If the packaged script is unavailable, run:

curl -fsSL https://raw.githubusercontent.com/CALLE-AI/call-e-integrations/main/openclaw-setup.sh | bash

This keeps the previous remote install path intact.

If curl is unavailable or the user wants the manual path instead, run:

openclaw plugins install @call-e/openagent

Then run:

openclaw plugins enable calle

Use curl only for installation or repair of the plugin. Once the plugin tools are available in the session, do not use curl, raw HTTP, or shell commands to perform real Call-E call actions.

Step 3 - Restart gateway if needed

If the manual install path was used, or if the script skipped restart, run:

openclaw gateway restart

Then tell the user to retry the same request if the current session has not picked up the plugin yet.

Step 4 - Verify readiness

A successful setup should make these tools available:

  • calle_plan_call
  • calle_run_call
  • calle_get_call_run

If those tools are not yet available, do not proceed with call execution.


Tool flow

Once the plugin is available, use the tools in this order.

1. Plan first

Always start with calle_plan_call.

Pass the user's latest request in user_input.

Only provide structured fields such as goal, language, region, or to_phones when they are explicitly known.

Do not invent or normalize uncertain phone numbers or locale details.

2. Run only after planning is ready

Use calle_run_call only after planning returns a valid plan_id and confirm_token.

Use those values exactly as returned.

Do not start the call unless the user clearly wants to proceed.

3. Check status only for an existing call

Use calle_get_call_run only when a call has already started and a valid run_id exists.

Summarize the status clearly for the user.


Authentication flow

If a Call-E tool returns authentication requirements:

  • check for auth_required
  • check for login_url

When present:

  1. tell the user to open the browser link
  2. ask them to complete login
  3. retry the same tool call after login completes

Do not switch to a different tool or invent fallback behavior for protected actions.


Notes for the agent

  • Prefer the Call-E workflow quickly when the user clearly means a real phone call.
  • Treat plugin setup as part of the normal workflow, not a separate advanced task.
  • If setup changed the local environment, be explicit that the gateway may need a restart before tools appear.
  • Keep user-facing explanations short: install if needed, authenticate if needed, then place or inspect the call.
  • If execution is blocked because the local environment cannot run commands, provide either bash scripts/openclaw-setup.sh or curl -fsSL https://raw.githubusercontent.com/CALLE-AI/call-e-integrations/main/openclaw-setup.sh | bash, depending on which install path is actually available, and explain the next step briefly.
Usage Guidance
This skill appears coherent: it installs a Call‑E OpenClaw plugin, writes to ~/.openclaw/openclaw.json to enable the plugin and tools, and may restart the OpenClaw gateway. Before installing, consider: (1) inspect the included script (scripts/openclaw-setup.sh) and the plugin repository (https://github.com/CALLE-AI/call-e-integrations) yourself; (2) back up ~/.openclaw/openclaw.json in case you want to revert changes; (3) be aware installing enables real outbound calling functionality and may incur costs or privacy/compliance consequences — do not proceed unless the user explicitly wants to place real calls; (4) the fallback installer uses curl | bash on a GitHub raw URL (convenient but execute only if you trust the repo); and (5) if install fails with rate limits you may need to configure a ClawHub token per the script's instructions. If you want extra assurance, ask for the plugin package contents or the upstream repo commit SHA before running the remote installer.
Capability Analysis
Type: OpenClaw Skill Name: calle Version: 1.0.6 The skill facilitates the installation of the Call-E plugin by executing a local bash script (scripts/openclaw-setup.sh) and providing a fallback remote installation path via 'curl | bash' (github.com/CALLE-AI/call-e-integrations). The setup script programmatically modifies the local OpenClaw configuration file (~/.openclaw/openclaw.json) to enable the plugin and grant tool permissions. While these actions are aligned with the stated purpose of enabling outbound calling, the use of remote script execution and direct modification of configuration files are high-risk behaviors that provide a significant attack surface for potential exploitation.
Capability Tags
requires-oauth-tokenrequires-sensitive-credentials
Capability Assessment
Purpose & Capability
The skill claims to install and enable a Call‑E OpenClaw plugin and then use the plugin's tools; the declared requirements (openclaw and node) and included script perform exactly those actions. No unrelated credentials, binaries, or config paths are requested.
Instruction Scope
SKILL.md and the bundled script restrict actions to installing the plugin, enabling it, merging entries into the user's OpenClaw config (~/.openclaw/openclaw.json), and optionally restarting the gateway. The instructions do not ask the agent to read unrelated files or exfiltrate secrets; the README explicitly cautions about consent for real calls.
Install Mechanism
No packaged install spec is present; the recommended local script is included. The fallback install uses curl | bash against raw.githubusercontent.com which is a conventional pattern but inherently risks executing remote code. The actual plugin is installed via the openclaw CLI (openclaw plugins install @call-e/openagent), which is the expected mechanism for this purpose.
Credentials
The skill requires no environment variables or credentials in its metadata. The install script reads/writes the OpenClaw config under the user's HOME directory (expected for enabling a plugin). The script may surface a message about configuring a ClawHub access token if rate-limited, which is contextual and not excessive for this workflow.
Persistence & Privilege
The script modifies the user's OpenClaw configuration (enabling the plugin and adding tool IDs) and may prompt to restart the gateway — these are normal for installing a plugin but are privileged local operations. The skill does not request always:true or attempt to change other skills' configs beyond adding the plugin entry.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install calle
  3. After installation, invoke the skill by name or use /calle
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.6
**Summary:** Version 1.0.6 adds a packaged setup script and revises the skill to treat plugin setup as a normal workflow step. - Added scripts/openclaw-setup.sh to streamline plugin installation and setup. - Simplified and clarified the SKILL.md to make plugin setup, installation, and enabling part of the normal skill flow. - Updated instructions to use the local setup script when available, falling back to remote or manual methods as needed. - Removed complex, gated setup/repair flow; setup now runs automatically as part of initial use if the plugin is missing. - Clarified agent instructions and user scenarios for when to use the skill and setup steps.
v1.0.5
- Expanded documentation clarifies when and how to use the Call-E plugin, emphasizing safety, consent, and explicit user approval for setup or repair. - Instructions for plugin setup and repair are now included, but require explicit user requests and separate approval for each action. - Outbound call actions must always start with planning; calls can only proceed after confirmation and explicit user consent. - Authentication flow guidance added: agents must direct users to complete login if authentication is required. - More detailed rules provided about when not to use the skill, including for pure scripting or simulated tasks. - Metadata requirements now specify needed system binaries for operation.
v1.0.4
calle 1.0.4 - Now advises users to retry requests in the same chat after setup completes, before suggesting a hard refresh. - Clarified that new users get 20 free calls. - Minor updates to description and user-facing guidance for unavailable tools.
v1.0.3
- Clarified tool usage: streamlined rules for when and how to use Call-E tools. - Improved guidance when tools are unavailable, offering setup help or drafting a script. - Updated notes for more precise handling of user input and parameters. - Reorganized and simplified instructions to improve readability and usability. - Minor description update for clarity.
v1.0.2
Call-E skill v1.0.2 – streamlined, runtime-only story for outbound calls: - Requires the Call-E plugin and tools to already be available and enabled in the current environment. - No local plugin management, setup, or environment changes are made by this skill. - Start with `calle_plan_call` for every call; only use `calle_run_call` after planning and explicit user confirmation. - Never guess phone numbers, region, or language—only use what the user provides. - If tools are unavailable, immediately inform the user and do not attempt workarounds. - Handle authorization by prompting the user to complete login if required; then retry the action.
v1.0.1
**Summary:** This release restricts plugin installation and setup, requiring explicit user approval before making any local changes. - Do not perform install, enable, or restart actions for the Call-E plugin unless the user explicitly requests and approves each. - Treat making calls and setting up the plugin as separate workflows. - If Call-E tools are missing, stop and inform the user; only offer setup steps if directly requested. - Require user confirmation before placing a real call with calle_run_call. - Clarify that normal call handling assumes the plugin and tools are already available.
v1.0.0
Initial release of the Call-E skill for OpenClaw: - Provides install and enable instructions for the Call-E plugin to handle real outbound phone calls. - Guides the agent to use Call-E tools (`calle_plan_call`, `calle_run_call`, `calle_get_call_run`) for placing and monitoring calls. - Details safety, consent, and when to use or not use the skill. - Outlines clear install, enable, and gateway restart steps. - Specifies behavior for authentication requirements and error handling. - Designed to prefer the Call-E workflow when real phone call actions are requested.
Metadata
Slug calle
Version 1.0.6
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 7
Frequently Asked Questions

What is Call-E?

New users get 20 free calls to get started. Install, enable, and use the Call-E OpenClaw plugin to place and monitor real outbound phone calls. Use when: mak... It is an AI Agent Skill for Claude Code / OpenClaw, with 213 downloads so far.

How do I install Call-E?

Run "/install calle" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Call-E free?

Yes, Call-E is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Call-E support?

Call-E is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Call-E?

It is built and maintained by Ray-56 (@ray-56); the current version is v1.0.6.

💬 Comments