ax-python-agent-context

Manage long-context memory in Python agents by selecting the correct Ax context strategy, such as context maps, context policies, or recall, to handle state and compaction without mixing incompatible patterns. The Skill guides you to the right mechanism based on your specific job.

The best way to choose between context maps, context policies, and recall is to evaluate your specific agent workflow needs, such as persistent corpus use or within-run compaction, and match them to the appropriate Ax Python mechanism for long-context management.

You need GEPA or ACE optimization for Python agent workflows when performing evaluation-driven offline optimization or tuning for agent systems, ensuring deterministic local checks and consistent behavior using the generated axllm package.

Yes, you can use the generated axllm Python API and examples for context optimization, which ensures syntax and behavior remain consistent with the AxIR-aligned runtime and optimizer APIs instead of relying on TypeScript-only calls.

Yes, choosing the correct context strategy requires the generated axllm package, package examples, and the AxIR-aligned runtime and optimizer APIs to maintain consistent syntax and behavior across Python agent workflows.