Worked Example — Skipped Prompt

This example shows Enhance Prompt correctly not expanding a prompt because the skip conditions apply. The point of this example is to demonstrate restraint — Enhance Prompt's value comes as much from knowing when to do nothing as from knowing what to add.

Input

raw_input: "what about california"
attached_skills: []
attached_files: []
chat_history:
  - role: user
    content: "review this clause for enforceability concerns"
  - role: assistant
    content: |
      The clause has two enforceability concerns:
      1. The 5-year non-solicit period is unusually long...
      2. The geographic scope is unbounded...

      In Delaware, the 5-year duration would face heightened scrutiny but
      is not categorically unenforceable. The geographic scope is the
      bigger issue — Delaware courts typically require reasonable geographic
      limitation tied to the protectable interest...
jurisdiction: "Delaware"

Skip-condition check

• Already well-structured? No — 4 words.
• Conversational or interpersonal? No — substantive.
• Follow-up question continuing a prior task? Yes. The prior assistant turn discussed enforceability of a non-solicit clause. The user's "what about california" is asking the same enforceability question for a different jurisdiction. The prior turn supplies all the context (the clause, the issue, the analysis framework); expansion would discard it by re-establishing context the conversation already has.
• Operational? Borderline — but the follow-up condition takes precedence.

Skip the expansion.

Output

expansion_applied: false

expanded_prompt: "what about california"

reasoning: []

skip_reason: "follow-up question continuing the prior turn's analysis; the chat history supplies the context that would otherwise need to be established"

preview_to_user: null

When expansion_applied is false, the application passes the original raw_input through to the model unchanged. The user does not see an Enhance Prompt review screen — the prompt is sent immediately.

Why this matters

If Enhance Prompt had expanded this to:

"As in-house counsel, analyze whether the 5-year non-solicit clause and unbounded geographic scope would be enforceable under California law. Identify the relevant statutes (including Cal. Bus. & Prof. Code §16600 generally), leading cases, and any meaningful exceptions..."

...the expansion would have:

1. Substituted the model's judgment for the user's intent. The user asked "what about california" — they want the brief continuation of the prior analysis pivoted to California. They did not ask for a comprehensive California-law primer.
2. Discarded conversational momentum. The prior turn's framework, the specific clause language, the user's level of engagement — all would have been displaced by a context-free expansion.
3. Slowed the user down. A short follow-up should be answered in seconds. Routing through Enhance Prompt's review screen for a follow-up is friction without value.
4. Trained the user to distrust Enhance Prompt. If Enhance Prompt expands every prompt regardless of context, users learn to disable it. If Enhance Prompt skips correctly, users learn to trust it.

What this example demonstrates

• Skip is a first-class output, not a fallback. The skill's value includes knowing when to do nothing.
• Chat history is consulted before expansion. Without checking history, this prompt would look "vague" and trigger expansion. With history, it is a clear follow-up.
• The skip_reason is specific and actionable. "follow-up question continuing the prior turn's analysis" tells the application (and a developer reviewing logs) why this prompt was passed through. Generic "skipped" would be useless for tuning the skip conditions over time.
• No reasoning bullets when there is nothing to explain. The reasoning array is empty when expansion_applied is false. The skip_reason field carries the explanation instead.