Automate content transfer when promoting captures to tasks/stories
Table of Contents
This page is a capture in the inbox bucket of the product backlog — a pre-sprint idea, not yet pulled into a sprint as a story.
What
Captures and tasks/stories are different document types with
different section schemas (a capture has * What=/=* Why=/
=* References=/=* See also; a task has * Goal=/=* Status=/
=* Acceptance=/=* Plan=/=* Notes=/etc.), so promoting a capture today
means scaffolding a fresh doc with =compass add task/story --id
<capture-id> (preserving the ID for traceability) and then manually
re-typing the capture's substance into the new doc's sections by
hand. Automate this: when --id matches an existing capture, have
compass read that capture's full body, pass it to the LLM with the
target schema, and have the LLM reformat/redistribute the content
into the new doc's sections in one step — then delete or archive the
source capture. The promoter should review/edit the result, not
retype it from scratch.
Why
Observed directly while promoting capture
EE237306-7782-4625-8880-47E8975530EA ("Scaffold model .org files
per entity type") into a task under the "Commission ores.qt.dq"
story: the ID was correctly preserved, but the capture's * What=/
=* Why=/=* References content had to be hand-copied and reworded
into the task's * Notes section because the two schemas don't
line up automatically. This is manual, error-prone (easy to drop a
reference or nuance in transcription), and repeats every single time
a capture is promoted — a purely mechanical step that an LLM-assisted
compass command could do reliably.
References
- Task Scaffold model .org files per entity type (regular/enum/junction) — the promotion that surfaced this gap.
projects/ores.compass/src/env_init.py=/=compass.py`add` command — where--idreuse and doc-type templating already live; this would extend that path.