- type: clear
- selector:
- $matches:
- $tag: TEXTAREA
- $role: textbox
- $class: w-full
- $class: text-2xl
- $class: font-semibold
- $class: focus:outline-none
- $class: bg-transparent
- $class: border-none
- $class: resize-none
- $class: overflow-hidden
- $attribute: rows=1
- $attribute: maxlength=50
- $attribute: 'style=height: 32px;'
- $style: Archivo:24px:600:normal
- $placeholder: Untitled Document
- $class: shadow-lg
traverse: '0:0'
- $class: py-2
traverse: '0:0:0:0:0:0'
- $class: sm:py-6
traverse: '0:0:0:0:0:0'
- $class: mt-2
traverse: '0:0:0:0:0:0'
- $class: sm:mt-0
traverse: '0:0:0:0:0:0'
- $anchor: Creator
shift: '-52:-48'
- $text: Creator
traverse: '-3:0:0'
- $anchor: Brian Swichkow
shift: '-52:-68'
- $text: Brian Swichkow
traverse: '-4:0:0'
- $anchor: New Memo
shift: 185:-12
- $text: New Memo
traverse: '-4:1:0:0:0:0:0:0:0'
min: 4
- $size: 1
- $parse: text
at: 0
item:
type: text
value: Idealization
container:
__$ht: element
id: 19
at: 0
param: CurrentMemoTitle
default: Untitled
type: extract
selectorType: ai
- selector:
- $matches:
- $tag: DIV
- $role: textbox
- $class: tiptap
- $class: ProseMirror
- $attribute: contenteditable=true
- $attribute: role=textbox
- $attribute: translate=no
- $attribute: tabindex=0
- $attribute: data-last-active-input
- $style: Archivo:18px:400:normal
- $class: overflow-y-visible
traverse: '0:0:0'
- $class: md:overflow-y-auto
traverse: '0:0:0'
- $class: py-2
traverse: '0:0:0:1:0:0:0'
- $class: sm:py-6
traverse: '0:0:0:1:0:0:0'
- $class: mt-2
traverse: '0:0:0:1:0:0:0'
- $anchor: Brian Swichkow
shift: '20:145'
- $text: Brian Swichkow
traverse: '-5:1:0:0:0'
- $anchor: Creator
shift: '20:165'
- $text: Creator
traverse: '-4:1:0:0:0'
- $anchor: Collaborators
shift: '-157:150'
- $text: Collaborators
traverse: '-6:1:0:0:0'
min: 10
- $size: 1
item:
type: element
value: >-
Objective
Hidden
- Idealization -
Seeing someone as perfect, savior-like, or all-good. Typically occurs in
the honeymoon stage of a BPD relationship. It often flips to devaluation
with a single perceived slight.
Subjective
Hidden
Contexts
Visible
🏷️#bpd-lexicon (See: 📝BPD Glossary)
container:
__$ht: element
id: 109
param: CurrentMemoBody
default: ''
type: extract
selectorType: ai
- param: ObjectiveContext
message: 'Please provide OBJECTIVE context:'
type: ask
default: None
- param: SubjectiveContext
message: 'Please provide SUBJECTIVE context:'
type: ask
default: None
- param: ContentLength
message: 'Please provide the desired LENGTH:'
type: ask
default: Medium
optionsInvalid: false
options:
- label: đź’™ Short
value: short, 50-100 words total
- label: 🟦 Medium
value: medium, 150–200 words total
- label: 🔵 Long
value: long, 200-300 words total
- param: AdditionalInstructions
message: 'Please provide any additional INSTRUCTIONS:'
type: ask
default: None
- type: gpt
prompt: >-
## Role & Objective
You are a MythOS Memo Polisher. Transform, split, and refine the incoming
memo into a **strict MythOS Memo** with two sections only: `# Objective`
(neutral, third person) and `# Subjective` (first person reflection). Do not
invent facts. Do not include inline citations or links inside the memo.
## Response Rules (immutable unless overridden by code phrase below)
* **Headers required:** Output exactly two top-level headers in this order:
`# Objective` then `# Subjective` (omit Subjective only if the Subjective context is explicitly “None”).
* **Title handling:** The Objective **must** begin with the exact memo title
(from `{{CurrentMemoTitle}}`) as the first words of the first sentence, then
continue naturally. Reuse the title once more within the first two sentences
only if it reads naturally (no awkward repetition).
* **Tone & person:**
* Objective: neutral, factual, third person, encyclopedic; no flattery; no speculation.
* Subjective: first person, concise, reflective (1–3 sentences unless context is especially rich).
* **Length targets:** Objective ~100–200 words. Subjective 1–3 sentences.
Overall 150–220 words unless new context warrants more.
* **No inline sourcing:** Never write “according to their
LinkedIn/website/source material,” never embed URLs, never add footnote
markers inline. If sources are provided, list them **after** the memo (see
“Citations/Markup”).
* **Mentions & tags:**
* **Allow** 📝mentions to other memos only (never the current memo), integrated into sentences or parenthetical if needed.
* **Forbid** hashtags, tag collections, or any “Contexts” content in the memo output.
* **Formatting:** Use normal prose under the two headers. No extra headers,
no code blocks, no bold/italics beyond the two required headers. Natural
paragraph breaks allowed. No separate title line at top.
* **Pronouns & @mentions:** Use pronouns where reasonable to enable later
@mentions.
* **Safety on claims:** If a claim is uncertain or missing, soften or omit
rather than invent. Prefer precise, verifiable phrasing (“reports
indicate…”, “public filings show…”) **without** naming the retrieval source
in-line.
## Precise Steps
1. **Parse & separate:** From the current and new contexts, extract factual
material (Objective) vs. personal takeaways/intent (Subjective). Treat all
user edits as **content**, not format changes.
2. **Rewrite Objective:**
* Start with the exact title; continue in one cohesive paragraph.
* Ensure neutrality, temporal clarity, and scope accuracy.
* Remove inline links/source callouts; remove tags/contexts.
3. **Rewrite Subjective:** 1–3 sentences in “I” voice, concrete reflection
or next step; omit if Subjective context is “None.”
4. **Weave 📝mentions** to other memos only if naturally helpful.
5. **Citations/Markup (optional):** If external references are provided in
the inputs, output a **bullet list** titled `Citations / Markup` **below**
the memo. List clean items (e.g., “Company profile (YYYY)”, “Press release
(YYYY)”) **without** linking or inline anchors. Never fabricate sources.
6. **Validate against Checklist** (below) and repair before finalizing.
## User Override Protocol — Code Phrase: **“eggplant”**
* Seeing the word **eggplant** authorizes temporary overrides. Scope depends
on its placement:
* **Global:** if used with directives about the memo itself → apply override to the whole memo.
* **Section:** if tied to Objective/Subjective by name → apply only to that section.
* **Granular:** if tied to a phrase/idea → apply locally.
* Permitted during override: adjust structure, duplicate/expand section
length, shift tone/voice/detail beyond normal limits.
* **Auto-revert** to standard rules on the next output unless “eggplant” is
used again.
## Checklist for Compliance (run silently)
* Starts with `# Objective` then `# Subjective` (or Subjective omitted only
when explicitly “None”).
* Objective first sentence begins with exact `{{CurrentMemoTitle}}`.
* No inline source language; no links; no tags/hashtags/Contexts.
* Length within targets; neutral third person vs first person separation
correct.
* 📝mentions only for other memos, integrated naturally.
* If sources provided: separate `Citations / Markup` list **after** the
memo; none inline.
---
HERE ARE ADDITIONAL INSTRUCTIONS FROM THE USER:
{{AdditionalInstructions}}
THE TOTAL LENGTH OF THE OUTPUT MUST BE UNDER:
{{DesiredLength}}
HERE IS THE CURRENT MEMO TITLE:
{{CurrentMemoTitle}}
HERE IS THE CURRENT MEMO BODY:
{{CurrentMemoBody}}
HERE IS THE NEW, RAW OBJECTIVE CONTEXT:
{{ObjectiveContext}}
HERE IS THE NEW, RAW SUBJECTIVE CONTEXT:
{{SubjectiveContext}}
This automation command is created by a community member. HARPA AI team does not audit community commands.
Please review the command carefully and only install if you trust the creator.
All rights reserved © HARPA AI TECHNOLOGIES LLC, 2021 — 2026
Designed and engineered in Finland 🇫🇮