CompSniper agent skill for AI coding agents

Install the official CompSniper skill so Claude Code, Cursor, Codex, and other skill-aware agents call the sold-comps API correctly on the first try.

The CompSniper agent skill teaches an AI coding agent everything it needs to build with the CompSniper API: endpoints, parameters, response fields, error handling, retries, and billing rules. Install it once and your agent writes working sold-comps code without you pasting documentation into the chat.

The skill follows the open Agent Skills format (a SKILL.md file with a short description plus reference files), so it works in any agent that supports skills.

Install with one command

Run this in your project or terminal:

npx skills add https://compsniper.com

The installer reads the public index at https://compsniper.com/.well-known/skills/index.json, downloads the compsniper skill, and places it where your agent looks for skills.

Download the files

Prefer to install by hand? Download the skill files directly:

Download SKILL.md api-reference.md examples.md

The skill has three files:

File	What it contains
`SKILL.md`	The core guide: `GET /v1/scrape` parameters and response, every other endpoint, error codes, and billing.
`references/api-reference.md`	The full endpoint reference, complete response field list, and the Mercari and Poshmark schemas.
`references/examples.md`	Worked examples in curl, JavaScript, and Python.

Manual install for Claude Code

Claude Code loads skills from a folder named after the skill. Keep the same layout as the download:

~/.claude/skills/compsniper/
  SKILL.md
  references/
    api-reference.md
    examples.md

Use ~/.claude/skills/ to make the skill available in every project, or .claude/skills/ inside a repository to share it with your team through version control. Start a new Claude Code session and the agent picks up the skill automatically whenever you ask about sold prices, comps, or resale value.

For other agents, place the same compsniper folder in that agent's skills directory.

What the agent learns

Authentication: every request sends Authorization: Bearer cs_.... Create a key in the dashboard.
Sold search: GET /v1/scrape with keyword, ebaySite, sold, count (up to 240), filters, and sorting, plus the summary block (median, mean, p25, p75, min, max).
Other marketplaces: Mercari (/v1/mercari) and Poshmark (/v1/poshmark/sold) with their own schemas.
High volume: bulk search for up to 20 keywords, card batch for up to 100 cards, and Max Mode for deeper research.
Errors and retries: which codes to retry (rate_limited, upstream_blocked, server_busy) and which not to (quota_exceeded, invalid_params). See errors.
Billing: only successful searches are charged, and X-Credit-Source: credits spends purchased credits directly.

Skill or MCP connector?

Use the skill when an agent is writing code that calls the CompSniper API with your API key, for example a pricing script, a reseller dashboard, or an inventory tool.

Use the MCP connector when you want your AI assistant to look up sold prices for you in the chat, with no code and no API key to manage.

Many developers use both: the connector for quick research and the skill for building.

Try it

After installing, ask your agent something like:

Write a Python script that gets the median sold price of a Nintendo Switch OLED on eBay US
from the last 30 days using CompSniper.

The agent reads the skill, calls GET /v1/scrape with the right parameters, and handles retries and errors the way the API expects.