Secondhand listing

Skill guycoful/secondhand-listing

Prepare Hebrew secondhand-item listings for Facebook Groups with fact-safe copy, optional Gemini image cleanup, joined-group research through the user's Chrome session, and human-approved drafting. Use when the user asks to sell, list, or publish a used item, including requests such as "פרסום פריט יד שנייה", while requiring the user to review and click the final Post button.From its SKILL.md

Install

npx -y skills add guycoful/secondhand-listing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

26 days oldThe repository was created 26 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.6 KB, ~1.4k tokens by cl100k_base, as published. Nobody here has run it

Secondhand Listing

Prepare a secondhand sale listing from item details through a ready-to-review Facebook Group draft.

Non-negotiable rules

Treat user-provided item details as the only source of truth. Never invent purchase date, original price, reason for sale, warranty, condition, scarcity, or flexibility.
Never click the final Facebook Post button. Stop after text and image are loaded so the user can review and publish personally.
Never send messages, call advertisers, contact buyers, join groups, or change account settings.
Search only for relevant groups the user already belongs to. Respect group rules and Facebook terms.
Never print, store in generated output, or commit API keys, cookies, passwords, or session data.
Keep the Hebrew copy natural and concise. Do not use an em dash. Put one space before question marks and exclamation marks.

Requirements

Python 3 with dependencies from requirements.txt
Chrome with an existing Facebook login
Chrome DevTools Protocol on port 9222 for the preferred browser flow
GEMINI_API_KEY in .env.secrets, .env, or the process environment for Gemini copy and image workflows

If Gemini is unavailable, use the keyless workflow with user-approved text and the original image.

Workflow

1. Collect facts

Ask for:

item type and brand
size or dimensions
exact condition
asking price in ILS
pickup city or area
additional facts the user wants disclosed
at least one source image when image handling is requested

If a missing detail would change the public claim, ask for it. Do not fill it in.

2. Create two listing variations

Run interactively:

python scripts/generate_listing.py

Or provide values directly:

python scripts/generate_listing.py \
  --brand "IKEA" \
  --size "120x80" \
  --condition "Good condition" \
  --price "250" \
  --location "Ramat Gan" \
  --description "Pickup only"

The script writes listing_draft.txt. Show both variations to the user and get a selection before drafting a post.

3. Clean the product image

When the user requests image cleanup and Gemini is configured, run:

python scripts/generate_image.py product.jpg -o catalog.jpg

The script must send this fixed prompt exactly:

תשנה את תמונת המוצר לתמונת קטלוג מקצועית בסגנון אמזון. דרישות: רקע לבן נקי (#FFFFFF) · תאורת סטודיו רכה ומפוזרת משני הצדדים עם צל עדין וטבעי מתחת למוצר · המוצר ממורכז ותופס 85% מהפריים · זווית 3/4 שמראה עומק · חדות מלאה בלי טשטוש · איכות קטלוג מסחרי פוטוריאליסטי · מראה נקי ומינימליסטי · שמירה על הצבעים המקוריים · סגנון מאקרו · מראה פרימיום מלוטש · 32k UHD · תשמור על אותו מוצר בדיוק עם אותם צבעים ופרטים ותשדרג רק רקע, תאורה וקומפוזיציה.

Check that the output preserves the product's shape, color, labels, damage, and identifying details. If it changes a material detail, use the original image.

If Gemini is unavailable or processing fails, keep the original image.

4. Research joined groups and prepare drafts

Prefer an already-open Chrome session launched with CDP:

chrome.exe --remote-debugging-port=9222

Then run:

python scripts/facebook_post.py --text-file listing_draft.txt --image catalog.jpg

The script asks for city, optional category, target groups, and listing variation. It loads the selected text and image in each group and pauses.

At every pause:

Tell the user which group is open.
Ask the user to verify the group rules, facts, price, image, and audience.
Let the user click Post personally.
Continue only after the user confirms.

The script waits 30 to 60 seconds between groups.

Keyless workflow

Use prepared content without Gemini:

python scripts/facebook_post.py --text "Your approved listing" --image path/to/photo.jpg

Or:

python scripts/facebook_post.py --text-file path/to/approved-draft.txt --image path/to/photo.jpg

CLI reference

`generate_listing.py`

--brand: item brand or type
--size: size or dimensions
--condition: exact condition
--price: asking price in ILS
--location: pickup location
--description: additional facts

`generate_image.py`

image: required source image path
-o, --output: output path, default catalog.jpg

`facebook_post.py`

--text: direct listing text
--text-file: approved draft path
--image: image path
--cdp: Chrome CDP port, default 9222
--profile: Chrome profile directory for the non-CDP fallback

Completion criteria

The task is complete only when:

the public facts match the user's input
the user selected or approved the copy
the chosen image was checked for material alterations
the draft is loaded in the intended group
the user retains the final publishing action

Do not claim that a post was published unless the user confirms the final click and the live post is visible.