pre-write-recall-checklisted
Install: claude install-skill yogsoth-ai/paper-reading
# Pre-Write Recall Check
Nugget-coverage method (adapted from TLDR/SciTLDR research): independently extract the paper's most important facts from raw source BEFORE looking at the bundle, then check bundle coverage. Catches omission errors — important content the bundle left out.
## Execution
Subagent — spawned via subagent-spawning/spawn-agent skill.
## Why Subagent
Independent nugget extraction must happen in a context that hasn't yet seen the bundle — reading the bundle first and then asking "did we miss anything?" biases the check toward confirming what's already there rather than discovering what's missing.
## Why This Is a Separate SOP From pre-write-precision-check
Precision checks whether stated claims are supported (commission errors); recall checks whether important content was left out (omission errors). These have different mechanisms (checking existing claims vs. independently generating a reference set) and different failure remedies (retry_deep_read vs. retry_deep_read_supplement per failure-routing) — see design spec §2 point 1. Do not collapse this sop's logic into pre-write-precision-check.
<!-- BEGIN available-tables (generated) -->
## Available SOPs
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use |
| --- | --- |
| spawn-agent | Spawn a customized CC subagent with full MCP tool access. |
| literature-research | Full-text paper reading via alphaxiv. Import from literature-engine. |
<!-- END available-tables (generated