Handoff

Use this skill to turn the current conversation and task state into a prompt that another agent can use to understand the context before waiting for the user's next instruction.

Principles

• Output a standalone prompt addressed to the next agent.
• Include only information needed to continue the work.
• Separate confirmed facts from assumptions and unresolved questions.
• Preserve user preferences, constraints, and safety rules that affect future actions.
• The handoff prompt must not instruct the next agent to start editing, running commands, or continuing implementation automatically.
• The next agent should only absorb the context and wait for the user's explicit next instruction.
• Do not include hidden chain-of-thought, internal deliberation, or irrelevant conversation history.
• Do not expose secrets, tokens, credentials, private keys, or sensitive file contents.
• If sensitive context matters, describe its existence and location generically instead of copying values.

Workflow