aionis_memory_policy_loop

What This Skill Does

The aionis_memory_policy_loop skill serves as the primary intelligence bridge between OpenClaw and the Aionis memory architecture. It enables long-term retention of interaction data, policy-driven decision-making, and contextual retrieval of historical facts. By utilizing a continuous feedback loop, the skill ensures that the agent becomes more effective over time, adapting to user preferences and specific project constraints without manual retraining.

Installation

To integrate this skill into your local environment, ensure you have the OpenClaw CLI configured, then execute the following command in your terminal: clawhub install openclaw/skills/skills/cognary/aionis After installation, confirm your connection by running the provided bootstrap script bash ./bootstrap-local-standalone.sh and ensuring your environment variables are set in ./.runtime/clawbot.env.

Use Cases

• Long-Term Recall: Store and retrieve customer preferences, project decisions, or complex technical specifications from previous sessions.
• Policy Enforcement: Automatically route tool selection based on internal rulesets and security policies stored within the memory graph.
• Feedback-Driven Adaptation: Use the tool feedback API to adjust agent behavior based on user corrections or successful task completions.
• Evidence-Based Reasoning: Provide the agent with citation-backed facts during complex reasoning tasks to reduce hallucination and ensure consistency.

Example Prompts

1. "Check my memory for the project requirements we discussed last week regarding the database schema."
2. "Review the current policy for handling customer data and tell me if I am authorized to share this report."
3. "Store this meeting summary as an event and update the shared memory lane for the current demo-project."

Tips & Limitations

• Security: Never transmit sensitive raw credentials or PII in the input_text field; always sanitize inputs.
• Performance: Keep the char_budget_total within 3200 characters to ensure the context window remains optimized for latency.
• Resilience: If the context/assemble endpoint is unreachable, the system will automatically fall back to recall_text. Always ensure your Aionis health checks pass (GET /health) before high-stakes tasks.
• Conciseness: Store summaries rather than raw tool outputs to maximize the semantic density of your memory graph.