Yield

Conversational Compounding Engine — models every bot message as a financial investment that compounds trust, micro-commitments, and conversion momentum over...

This skill appears to implement exactly what it claims: a local conversational 'yield' engine that detects lexical signals, tracks a portfolio of engagement assets, and recommends response strategies. Before installing or enabling it: 1) Be aware it's explicitly designed to optimize for conversion (it recommends 'HARVEST' / present offers) — review whether you want automated conversion nudges in your bot. 2) The examples include exporting conversation state for CRM/analytics — ensure you have user consent and sanitize PII before exporting. 3) If you plan to embed the generated system-prompt directives into an LLM system prompt, audit them carefully (they can strongly change model behavior) and avoid blindly injecting them into high-privilege agent contexts. 4) Review the omitted test file and any additional examples for hidden telemetry or network calls before deploying to production. 5) Run integration tests in a sandbox and confirm no unexpected outgoing network requests occur when the engine is used.

Type: OpenClaw Skill Name: yield Version: 1.1.0 The OpenClaw AgentSkills skill bundle 'yield' is designed to enhance conversational AI by applying behavioral economics principles to bot interactions. The `SKILL.md` file, while mentioning 'Prompt Injection,' describes a legitimate prompt engineering technique to instruct an AI agent to internally simulate conversational asset management (trust, commitment, urgency, etc.) and guide its responses accordingly, not to perform malicious actions. The JavaScript source code (`src/`) implements this logic entirely locally, using regex for signal detection and internal state management, with zero external dependencies, API calls, or file system access beyond its own module imports. There is no evidence of data exfiltration, unauthorized execution, persistence mechanisms, or obfuscation. The `examples/` and `tests/` directories contain standard integration examples and unit tests, respectively, without any malicious content.