pushbullet-automation
by ComposioHQpushbullet-automation helps agents automate Pushbullet tasks through Composio Rube MCP by discovering current tool schemas, checking the Pushbullet connection, and executing notification or device workflows safely.
This skill scores 68/100, which makes it acceptable but limited for directory listing. Directory users can understand when to use it and how an agent should begin Pushbullet automation through Rube MCP, but they should expect a runtime-discovery-driven wrapper rather than a self-contained Pushbullet playbook with concrete schemas or bundled examples.
- Clear trigger and scope: it is specifically for automating Pushbullet operations through Composio's Pushbullet toolkit via Rube MCP.
- Provides operational prerequisites and setup steps, including adding the Rube MCP endpoint and activating the Pushbullet connection with RUBE_MANAGE_CONNECTIONS.
- Includes an agent-friendly workflow pattern that repeatedly instructs the agent to call RUBE_SEARCH_TOOLS first for current schemas and execution plans.
- Execution depends on an external Rube MCP server and an active Pushbullet connection; the repository does not include scripts, local tests, or bundled reference files.
- The skill intentionally defers concrete tool names and schemas to runtime discovery, so users get less task-specific detail than a fully documented Pushbullet workflow skill.
Overview of pushbullet-automation skill
What pushbullet-automation does
pushbullet-automation is a Claude skill for automating Pushbullet actions through Composio’s Rube MCP. Instead of guessing Pushbullet API shapes, the skill instructs the agent to discover the current Rube tool schemas first, verify the Pushbullet connection, then execute the requested notification, device, or messaging workflow with the right tool inputs.
Best fit for Workflow Automation users
This skill is useful if you already use Pushbullet for cross-device alerts and want an agent to trigger actions as part of a broader workflow: sending yourself a link, pushing a note to a device, notifying after a job completes, or wiring Pushbullet into a personal automation loop. It is especially relevant for users building agentic Workflow Automation where Pushbullet is one channel among several tools.
Main adoption requirement
The key dependency is not the skill file itself; it is Rube MCP access plus an active Pushbullet connection. The skill requires RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS. If your Claude or MCP client cannot connect to https://rube.app/mcp, or you cannot authorize Pushbullet through Composio, the pushbullet-automation skill will not be executable.
What makes it different from a generic prompt
A generic prompt may ask the model to “send a Pushbullet message,” but it may invent tool names or stale parameters. This skill’s main differentiator is its workflow discipline: search tools first, use returned schemas, check the Pushbullet connection, and only then call the appropriate Rube action. That lowers the risk of failed calls caused by outdated assumptions.
How to Use pushbullet-automation skill
pushbullet-automation install and setup path
Install the skill in a compatible skills environment, for example:
npx skills add ComposioHQ/awesome-claude-skills --skill pushbullet-automation
Then add Rube MCP to your client configuration using:
https://rube.app/mcp
Before expecting any Pushbullet action to work, confirm that RUBE_SEARCH_TOOLS is available. Next, use RUBE_MANAGE_CONNECTIONS with toolkit pushbullet. If the returned status is not ACTIVE, complete the authorization link and re-check the connection before running the automation.
Inputs the skill needs from you
For good pushbullet-automation usage, provide the agent with the actual outcome, not just the app name. Include:
- Action type: note, link, file-like notification, device-targeted push, or general alert
- Recipient scope: all devices, a specific device, or your own account
- Message content: title, body, URL, or alert text
- Timing context: send now, after another task, or only if a condition is met
- Safety rules: ask before sending, avoid duplicate alerts, or summarize before execution
Weak prompt: “Use Pushbullet to notify me.”
Stronger prompt: “Use pushbullet-automation to send a Pushbullet note to all my devices after you finish the report. Title: Report ready. Body: include the output file path and a one-sentence summary. First verify the Pushbullet connection and discover the current Rube schema.”
Recommended execution workflow
A reliable pushbullet-automation guide should follow this sequence:
- Ask the agent to call
RUBE_SEARCH_TOOLSfor the specific Pushbullet task. - Have it inspect returned tool slugs, required fields, and warnings.
- Confirm the Pushbullet connection with
RUBE_MANAGE_CONNECTIONS. - Prepare the exact tool payload from your requested title, body, URL, device, or condition.
- Execute once, then report whether the push succeeded and what was sent.
This order matters because Rube tool schemas can change. The skill explicitly tells the agent not to rely on memory when a live schema lookup is available.
Repository files to read first
This skill is intentionally compact. Start with composio-skills/pushbullet-automation/SKILL.md; there are no extra README.md, scripts/, rules/, resources/, or references/ folders in the current file tree. Pay close attention to the prerequisites, setup notes, and tool discovery examples because they contain the operational value of the skill.
pushbullet-automation skill FAQ
Is pushbullet-automation beginner friendly?
It is beginner friendly if your client already supports MCP and you are comfortable following an authorization link. It is not a one-click Pushbullet installer. The main beginner hurdle is understanding that Claude does not talk to Pushbullet directly; it uses Rube MCP, and Rube must have an active Pushbullet connection.
When should I not use this skill?
Do not use pushbullet-automation if you only need to draft notification text without sending it, if your environment cannot run MCP tools, or if your organization blocks third-party workflow connectors. Also avoid it for high-risk or regulated alerts unless you add confirmation steps, logging, and clear human approval rules.
How does it compare with direct Pushbullet API use?
Direct API use may be better for production software, custom retry logic, or strict infrastructure control. The pushbullet-automation skill is better for agent-driven workflows where Claude needs to discover available actions, assemble valid inputs, and perform Pushbullet tasks without you writing integration code.
Does the skill know every Pushbullet parameter?
No. The skill is designed around live tool discovery, not hardcoded certainty. For current fields, device identifiers, required payloads, and execution plans, the agent should call RUBE_SEARCH_TOOLS first. This is a feature: it reduces reliance on stale documentation or model memory.
How to Improve pushbullet-automation skill
Improve prompts with execution constraints
The easiest way to improve pushbullet-automation results is to specify both the message and the operating rule. For example: “Send only one Pushbullet alert if the deployment fails; include service name, environment, failure reason, and next step; ask me before sending if the message contains secrets.” This gives the agent enough context to avoid noisy or unsafe pushes.
Reduce common failure modes
Common failures include skipping tool discovery, trying to send before the Pushbullet connection is active, omitting required fields, or targeting the wrong device. Prevent these by requiring the agent to show the discovered tool name and planned payload before execution when the notification matters.
Iterate after the first output
After a first successful push, refine the workflow rather than rewriting it. Ask for shorter titles, richer bodies, device-specific routing, deduplication rules, or conditional sending. Example: “Next time, only push to my phone, limit the body to 300 characters, and include the GitHub Actions run URL.”
Extend pushbullet-automation for stronger workflows
For more advanced Workflow Automation, pair pushbullet-automation with upstream tasks such as monitoring, report generation, deployment checks, or calendar reminders. The skill becomes more valuable when Pushbullet is the delivery endpoint for a larger agent workflow, not just a standalone “send message” command.