daily-note-addition

name	daily-note-addition
description	Park one or more loose ideas into today's daily note as flat checkbox captures. Use when the user has thoughts — a new todo, a research angle, a thing to remember — they want filed under today's execution plan rather than triaged to the inbox or the current project note. This is the lightweight flat-capture path, distinct from `/breakdown-tasks`.

Objective

Pin one or more ideas to today's daily note as actionable checkboxes under the bucket that best fits each idea's domain. Stay terse — this is a capture tool, not a planning session.

Workflow

1. Locate the note

• Open 10_Daily/YYYY-MM-DD.md for today.
• Missing? Fall back to the most recently modified file in 10_Daily/ (glob + sort by mtime desc).
• Read it once and extract the existing bucket headings (e.g. **a1. Thesis & Paper Work**, **c1. Odd Jobs**). Buckets vary per day — never hard-code them.

2. Split the input

• One bullet per idea. Split on blank lines or sentence breaks when the user gives several thoughts at once.
• Preserve the user's wording; fix only obvious typos. Do not paraphrase or expand — this is capture, not rewriting.

3. Classify each idea

You already read the note in Step 1 — use the actual bucket headings and existing bullets inside them as your classification signal. Match each idea to the bucket whose headline and contents are the closest semantic fit. Trust the note, not a fixed table: bucket labels, focus, and ordering shift from day to day.

If no bucket is a clean fit, append the idea to the last bucket in the Priorities section and flag the weak match in Step 6's report so the user can move it.

4. Format the bullet

- [ ] <idea text> (~ X mins)

• Estimate: rough ~ X mins (nearest 5m) when scope is obvious; (~?) otherwise. Reason contextually, never from keywords.
• Wikilink: when an idea clearly ties to a project already referenced in the note's ## Related Projects section (or a project wikilinked elsewhere in the note), weave that [[ProjectName]] into the bullet text naturally — not tacked on at the end. Pull the project names from the note you just read, not from memory.

5. Insert

For each bullet, find where the chosen bucket's content ends in the note you read in Step 1 — wherever its items give way to a clearly different section. Derive that boundary from the note's actual structure, not a fixed marker — the same principle as Step 3 classification. Append the new bullet as the bucket's final item using Edit with a multi-line anchor drawn from the note's real surrounding content; if the bucket has nested sub-checkboxes, the anchor should span the last top-level item through the closing boundary so the match is unique in the file.

6. Report

Terse confirmation only: target note date (flag if it wasn't today), per-idea → bucket mapping, bullet count. Do not echo the full note.

Edge cases

• Semantic duplicate already present: if an existing bullet in the note overlaps meaningfully with a new idea — same object + verb + scope, not just shared keywords — skip that idea and tell the user which bullet it collided with. Err toward flagging; the user can override on the next turn.
• Empty input: ask for the idea(s) via AskUserQuestion.
• No daily notes at all: tell the user to run /start-my-day first.