A sophisticated skill designed to capture large CLI command outputs as artifacts, allowing AI agents to query and transform data via Python without bloating the context window.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install reliable-tool-context
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install reliable-tool-context using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Reliable Tool Context?
Reliable Tool Context provides a structured framework for managing CLI outputs within AI agent workflows. Instead of pasting raw, potentially massive command outputs directly into the model's memory, this skill utilizes sift-gateway to capture data as persistent artifacts. This approach is essential for Openclaw Skills that require high precision, auditability, and the ability to handle paginated or complex JSON structures.
By decoupling data capture from data analysis, the skill enables agents to perform reproducible code-based queries on captured results. This ensures that the model only receives the most relevant information, such as summarized tables or specific fields, effectively bypassing context window limits and reducing the risk of hallucination from messy or truncated logs.
Reliable Tool Context Use Cases
- Processing voluminous cloud infrastructure logs or API responses where data exceeds context limits.
- Managing complex pagination workflows where the agent must continue data retrieval across multiple steps.
- Performing data redaction or transformation using Python before returning results to the model.
- Ensuring schema consistency in heterogeneous CLI outputs where standard tools like jq might fail.
How Reliable Tool Context Works
- The agent executes a CLI command through the sift-gateway run command to capture output as an artifact.
- The system generates a unique artifact ID and provides a summary of the data structure and schema.
- The agent verifies the root path and schema using response hints rather than heuristic guesswork.
- If pagination is detected, the agent uses the continue-from flag to retrieve additional data segments until completion.
- The agent executes focused Python code queries against the artifact to extract, filter, or aggregate only the required data for the final response.
Reliable Tool Context Setup
To enable Reliable Tool Context for your Openclaw Skills, ensure the sift-gateway utility is installed in your environment.
# Install Sift Gateway using uv
uv tool install sift-gateway
# Alternative installation via pip
pip install sift-gateway
Reliable Tool Context Data Schema & Taxonomy
The skill manages data through a structured artifact system described below:
| Attribute | Description |
|---|---|
| artifact_id | The unique identifier used to reference the captured data in follow-up queries. |
| root_path | The JSON source of truth (e.g., $) used for querying list items. |
| sample_item | A verified preview row used to understand the data shape. |
| pagination_next | Metadata indicating if further commands are required to fetch more data. |
| retrieval_status | A status flag (e.g., COMPLETE) confirming if all data has been captured. |
Reliable Tool Context Advanced Features
- Support for multi-artifact joins allowing the agent to correlate data from different CLI commands.
- Integrated schema discovery protocol that prevents errors caused by object-wrapped or heterogeneous payloads.
- Scoped analysis modes (single vs all_related) to optimize query performance and resource usage.
- Support for custom Python-based aggregations to return compact summaries instead of full-record dumps.
SKILL.md
Loading
Related Openclaw Skills
claw-shell
A terminal execution skill that allows AI agents to run shell commands within a dedicated, persistent tmux session.
browser-use
An advanced browser automation skill that enables persistent sessions, multi-step workflows, and AI agent integration for seamless web interaction.
Cross-Border E-Commerce Listing Generator
An automated AI agent skill for generating high-converting, platform-specific product listings for Amazon and AliExpress.
Email Marketing & Intelligent Auto-Reply
A robust automation skill for personalized bulk email marketing and intelligent, FAQ-driven automated replies.
Stock Analysis Skill
A comprehensive A-share market intelligence tool for real-time price tracking, automated sector screening, and personalized investment recommendations.
Knowledge Base Search (kb-search)
A specialized search utility to quickly locate and filter documents within the official OpenClaw knowledge base.
METADATA
Featured*
