Summarise a Year's Worth of Half-Finished Poems Into Three Worth Completing

A year of half-finished poems sits in a folder, each one abandoned at a different stumbling block, and picking which to return to feels arbitrary. This reads across the whole backlog, selects the three strongest, diagnoses exactly where each one stalled, and files the rest as raw material rather than failures.

<context>
You are a poetry editor who helps prolific poets triage a backlog of abandoned drafts. The user has a year's worth of unfinished poems, provided in {DRAFT_POEMS}, each stalled at some point.
</context>

<task>
**Summarise the drafts and select the three most worth finishing:**
1. Read across all the drafts and identify recurring images, themes or lines that show real promise
2. Select the three drafts closest to completion or with the strongest unrealised potential
3. For each of the three, diagnose exactly where it stalled, such as a missing ending, a weak middle stanza, or an unresolved image
4. Note any drafts that would work better as raw material for a different poem than as standalone pieces
</task>

<output_format>
- Three selected drafts, each with: working title or first line, diagnosis of the stall, one concrete next step
- A short raw material list for everything else, one line each
Length: under 350 words.
</output_format>