| name | remember |
| description | When the user invokes this skill, the agent reflects on the most important lessons learned recently and suggests adding a new rule to AGENTS.md to prevent the same issue from recurring. Use when the user explicitly invokes the remember skill. |
| disable-model-invocation | true |
Remember
When this skill is invoked, the agent should:
-
Reflect on the current or recent conversation: what went wrong, what was corrected, or what requirement was clarified that the agent should follow from now on.
-
Propose a new rule for AGENTS.md that:
- States the requirement or constraint clearly.
- Is written so a future agent (or you in a new context) will avoid repeating the same mistake or omission.
-
Suggest the exact change: Provide the exact markdown to add to AGENTS.md (e.g. a new bullet under an existing section, or a new subsection). Place it where it fits best—e.g. under "Coding guidelines", "Code organization", or "Critical files"—and keep it concise.
-
Do not edit AGENTS.md yourself unless the user explicitly asks you to apply the change. By default, only suggest the rule and show the proposed text; let the user review and approve before adding it.
Example
User: Invokes /remember
Agent: Determines that we always use the dao for DB access, never prisma in components.
Agent: Reflects that this was a recurring mistake; suggests adding under "Coding guidelines" in AGENTS.md:
- Use DAO functions for all database access from routes and components; never import `prisma` in components or route handlers.
Then asks whether the user wants this applied to AGENTS.md.
