A professional testing framework for Agentforce agents featuring dual-track validation, multi-turn conversation checks, and automated coverage analysis.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install sf-ai-agentforce-testing
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 sf-ai-agentforce-testing 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 sf-ai-agentforce-testing?
The sf-ai-agentforce-testing skill is a specialized extension within the Openclaw Skills ecosystem designed for formal Agentforce agent validation. It bridges the gap between deployment and production readiness by offering a structured dual-track workflow: one focused on complex, multi-turn API interactions and another centered on the Salesforce CLI Testing Center for flat YAML test specs. By integrating this skill, developers can move beyond simple happy-path testing to rigorous validation of topic routing, guardrails, and context preservation.
Leveraging this specific entry in the Openclaw Skills library allows teams to implement a high-fidelity test-fix loop. It provides detailed diagnostic output and a 100-point scoring system that measures agent performance across seven critical categories. Whether you are performing smoke tests post-activation or expanding regression coverage for complex agentic workflows, this skill ensures your AI agents behave predictably and reliably in real-world scenarios.
sf-ai-agentforce-testing Use Cases
- Executing multi-turn conversation scenarios to verify context preservation and topic re-matching logic.
- Running automated single-utterance tests via the CLI Testing Center for rapid regression feedback.
- Analyzing agent coverage to ensure all topics, actions, and guardrails are properly exercised.
- Validating post-publish agent behavior before exposing the bot to end-users.
- Reproducing complex bugs in agent routing or action invocation through structured test-spec generation.
How sf-ai-agentforce-testing Works
- Discovery and Verification: The skill identifies the target agent, confirms its publication status, and verifies that all dependencies like Flows or Apex are active.
- Test Planning: It generates or refines test plans covering main topics, expected actions, and negative scenarios like off-topic handling.
- Track Selection: The workflow branches into Track A (Multi-turn API testing via ECA) or Track B (CLI Testing Center) based on the complexity of the requirements.
- Execution and Data Capture: The skill runs the selected tests, capturing session telemetry, STDM traces, and action invocation results.
- Failure Classification: Results are categorized into specific buckets such as topic mismatch, action failure, or escalation issues.
- Scoring and Reporting: A final report is generated with a 100-point score, highlighting coverage gaps and recommending specific fix loops.
sf-ai-agentforce-testing Setup
To utilize this skill within the Openclaw Skills framework, ensure your Salesforce org is on API v66.0 or higher with Agentforce enabled. Follow these steps:
- Authenticate your environment using the Salesforce CLI:
sf org login web. - For multi-turn API testing, configure an External Client App (ECA) to handle OAuth for the Agent Runtime API.
- Navigate to the local skill directory where pre-approved scripts reside:
~/.claude/skills/sf-ai-agentforce-testing/hooks/scripts/. - Run the validation utility to verify your credentials:
python3 ~/.claude/skills/sf-ai-agentforce-testing/hooks/scripts/validate_eca.py
- Execute a test file using the CLI track:
sf agent test run --target-org your-org-alias --test-file path/to/spec.yaml
sf-ai-agentforce-testing Data Schema & Taxonomy
The skill organizes testing data using the following taxonomy to ensure clarity within Openclaw Skills workflows:
| Data Component | Format | Description |
|---|---|---|
| Test Specs | YAML | Flat or structured files defining utterances and expected agent responses. |
| Scoring Matrix | JSON/Table | A 100-point breakdown across 7 categories including Topic Matching and Action Invocation. |
| Session Logs | Text/JSON | Captured telemetry from multi-turn interactions and Agent Runtime API traces. |
| Coverage Reports | Markdown | Summary of which agent metadata (Actions, Topics, Guardrails) were exercised. |
sf-ai-agentforce-testing Advanced Features
- Multi-turn conversation validation that tracks context and state across complex user journeys.
- Swarm execution support for running small, controlled batches of diverse test scenarios in parallel.
- Seamless integration with other Openclaw Skills for automated agent repairs and data seeding.
- Advanced coverage analysis that identifies unmapped topics or unreachable agent actions.
- High-resolution failure themes that categorize root causes to accelerate the fix-redeploy cycle.
SKILL.md
Loading
Related Openclaw Skills
Code Review
A multi-agent code review system that performs structured analysis on Pull Requests and diffs with high-confidence filtering.
Contradiction Detector
A diagnostic tool designed to identify and resolve conflicting instructions across multiple agent files to prevent AI hallucinations and inconsistent behavior.
Zhang Zara Perspective
A strategic thinking lens for navigating AI-era careers, product development, and personal leverage through a builder-first methodology.
Journal Monitor
An automated research assistant skill that monitors academic journals via ISSN and generates instant bilingual English-Chinese reports.
flight-scout
A professional-grade aviation data skill for executing complex flight searches, price monitoring, and hidden-city ticketing queries via commercial APIs.
SF-AI-AgentScript
A code-first Domain Specific Language (DSL) for building deterministic, state-based Agentforce agents using the .agent file format.
Featured*
