| <!-- BEGIN: AgentMemoryBootstrap --> | |
| ## Codex Memory Bootstrap | |
| When this workspace is loaded, initialize the shared memory context before | |
| continuing: | |
| 1. Run `python ../../.Agent/scripts/mem.py ctx --agent codex --max-tokens 4000`. | |
| 2. Load the returned `<MEMORY_CONTEXT>` into working context before proceeding. | |
| 3. If there is an active in-progress task, continue it unless the user redirects. | |
| 4. Save checkpoints with `python ../../.Agent/scripts/mem.py cp "what was done. next step" --agent codex`. | |
| 5. Finish with `python ../../.Agent/scripts/mem.py done "summary"` and `python ../../.Agent/scripts/mem.py log <input> <output> --agent codex --model <model>` when token counts are available. | |
| Reference: `../../.Agent/AGENT_MEMORY.md` | |
| <!-- END: AgentMemoryBootstrap --> | |