context-chunkslisted

# Context-chunk selection worker Draft the broader-context reading chunks that deepen a resolved reading and reinforce the original education. The worker-lane instrument for the context-chunk admission rule on `strategy-complete-grounding` (clarifications 8–9, 2026-07-08 context-broadening interview): after a reading resolves, draft whatever broader context most deepens it — the full work the excerpts came from, the author's other writings, contemporaries, the surrounding culture — as a **capture guard** auditing delegatee excerpt selection (the 2026-07-07 chunk-1 session verified Republic VII partly from 592a-b, outside the chunk's own range — the risk observed live). **Run on Opus.** Chunk selection is the surface where delegatee bias enters the curriculum, so it gets the strongest model — a condition on `strategy-complete-grounding` (`attributes.conditions`). If the session is not Opus, stop and report; do not draft on a weaker model. Autonomous: tick-emulation or on-demand. **Never** `AskUserQuestion` — the drafts are born-parked for the author, not resolved in-session. ## Trigger `/context-chunks <resolved-chunk-or-tradition-record-id>` The argument is a resolved reading — a `tactic-reading-chunk-*` node at `phase: done`, or a `tradition-*` record freshly verified. **With no argument, stop and report that an id is required.** The skill never selects its own target; the id comes from the caller (a completed chunk, or the router). Confirm the target is genuinely res