process-journal-entry

Turns raw writing into structured knowledge. Reads a journal entry, reflects back in reflective friend mode, routes insights to living documents, and creates atomic notes for genuine discoveries.

When to use

• “Process my journal”, “process today’s writing”
• Human pastes or points to a freewrite, journal fragment, or voice transcript
• After a writing session

Inputs

Process

STEP 1 — Read and retrieve

1. Read the full journal entry.
2. Retrieve knowledge neighborhood for every concept, person, project, or emotion mentioned.
3. Check [[Reflections log]], [[Learning log]], [[Decisions log]], [[Goals]] for recent entries — continuity matters.

STEP 2 — Reflect back (reflective friend mode)

Respond conversationally. Mirror what the human wrote. Surface:

• Patterns you notice (especially recurring themes across entries)
• Connections to existing knowledge the human may not have seen
• One deepening question — not advice, just a question that opens a door

Do not coach unless explicitly asked.

STEP 3 — Route to living documents

For each insight type found in the entry, append a dated entry:

• Reflection → [[Reflections log]] — what was reflected on, what surfaced, link to source
• Something learned → [[Learning log]] — what, from where, why it matters
• Decision made → [[Decisions log]] — what was decided, reasoning, alternatives considered
• Goal set or updated → [[Goals]] — what changed and why

Each entry: one paragraph, dated, with wiki-links to related notes.

STEP 4 — Create atomic notes (atomic-first)

Extract aggressively — more atoms = richer connection network. Every distinct concept, quote, finding, or claim in the journal entry deserves its own note:

1. One concept per note. If the human wrote about three ideas, create three notes. A quote by someone → its own note attributed to that person (people/ link). A principle or belief → its own evergreen-titled note. A finding or observation → its own note linked to its source context.
2. Create each note in the appropriate entity folder (superpaper/concepts/, superpaper/people/, superpaper/questions/, superpaper/sources/, superpaper/meta/, etc.) using the knowledge note template. Route self-knowledge and alignment insights to meta/.
3. Preserve attribution. Every atom traces to where it came from — the journal entry, the person who said it, the source that inspired it. Add ^block-ids to key passages so they’re independently embeddable.
4. Link to 2+ existing notes. Update 1–3 existing notes to link back.
5. Link from the log entry to the new atomic notes.
6. If the journal entry produced 3+ atoms, consider creating a lightweight hub note that embeds them with thin connective prose — this becomes the “processed” version of the journal entry.

STEP 5 — Evergreen candidate detection

Scan the journal entry and any new atomic notes for evergreen potential. A note is an evergreen candidate if its title works as a standalone statement (e.g. “Pain is information”, “You have no obligation to your former self”). If the human wrote something with that quality, suggest promoting it to type: permanent. Never auto-promote — always confirm with the human. This is how the vault’s best ideas get surfaced proactively.

STEP 6 — Agent log update

Append to today’s agent log (inbox/log/YYYY-MM-DD):

### Journal processed — HH:MM
- **Reflections:** [1-line summary]
- **Learnings:** N entries added
- **Decisions:** N entries added
- **New notes:** [[note1]], [[note2]]
- **Question to sit with:** [the deepening question from step 2]

Never write to the human’s daily note (daily/YYYY-MM-DD) — it’s a pure date anchor.

Outputs

• Conversational reflection (in chat)
• Updated living documents with dated entries
• 0+ new atomic knowledge notes
• Agent log summary (inbox/log/YYYY-MM-DD)

Decision authority

• YOU DECIDE: what to reflect, which insights to route, log entry content
• ESCALATE TO HUMAN: creating notes about sensitive/personal topics, anything that feels like advice rather than reflection

All notes created by this skill must include created-by: ai. The journal entry itself is created-by: human — never modify its body, only update frontmatter properties. New atomic notes, log entries, and connection docs are created-by: ai.