search.peptides

Every research-peptide vendor and listing, compared: ~20,000 listings from ~240 vendors across ~770 peptides, ranked by USD per mg, with independent lab results for ~460 vendors and each vendor's contact details (email, phone, WhatsApp, Telegram, socials, websites).

• Web: https://search.peptides.pit.world (on Moshpit DNS: https://search.peptides)

• Faceted search, the r4ck.dev/servers rail: peptide, price per mg, price, vial size, lab testing, verification score, Finnrick rating, payment, ships from, form, single or blend, how you order, vendor. Every count is exact and every option is a link.

• API, no key: /api/v1/search, /api/v1/vendors, /api/v1/vendors/:slug, /api/v1/peptides, /api/v1/peptides/:slug, /api/v1/rules, /api/v1/export

• MCP: POST /mcp or bunx searchpeptides mcp (tools search_listings, search_vendors, get_vendor, get_peptide, list_peptides, get_rules)

• CLI and TUI: bunx searchpeptides search "retatrutide lab tested under $1/mg", … vendors --contact email, … vendor <slug>, … rules, … tui

• Mirrored to https://nichedb.dev/c/peptides

Verification score

Every vendor gets a 0-100 verification score computed in src/score.js from evidence that can be checked: Finnrick's per-test records (pass/fail, lab named, certificate on file with a public record page, lot/batch identifier, who sent the sample), Finnrick's own rating, and the vendor's published terms. Price, reviews and certificates a vendor posts itself are not inputs. Fewer than 3 classifiable tests in two years shows "insufficient data", none shows "unscored"; nothing is guessed. Weights, rules and every score change: /rules (/api/v1/rules, peptides rules, MCP get_rules). The score is recomputed after every daemon update and every 6 h; each change goes to score_log with its reason. Change a rule: bump RULES_VERSION and add a RULES_HISTORY entry.

Scores and test counts come from Finnrick's independent lab testing; COA findings are from documents the vendor publishes. Not affiliated with Finnrick or any vendor. For research use; not medical advice.

Related MCP server: Peptide Library MCP Server

Data

Kept fresh by the updater daemon, @profullstack/scraper-daemon (the same package ch4nnels.com, r4ck.dev, rssamplifier.com and p0dcasters.com run), in src/daemon.js, as the worker role of the one process. State lives in the same SQLite file; /api/v1/daemon shows it.

Fetching goes through the package: robots.txt, one request at a time per host, public addresses only, every redirect re-checked, 429 holds the host. peptides daemon run <task> [vendor] runs one task by hand; data/snapshot.json.gz seeds a fresh install offline.

Run

bun install
bun src/main.js            # :8802, loads the snapshot on first boot, then syncs and enriches
bun test

Deploy (dev2)

/opt/search.peptides, unit search-peptides, port 8802, bin/install.sh (OpenInstall). app.env at /root/.local/share/search-peptides/app.env from vault profullstack/search-peptides--prod. Every merge to main deploys through the forced command deploy/search-peptides-deploy; deploy/after-checkout.sh takes an online copy of the database first (~anthony/Backups/root-db/search-peptides/search-peptides-predeploy-*.sqlite.gz, newest 5). Moshpit origin vhost: deploy/nginx-search.peptides.conf, installed as /etc/moshpit/templates/search.peptides.conf.

Sponsored placement ($49 / 30 days) is the one thing for sale; sponsored vendors are always labelled. It is paid through a CoinPay hosted checkout (any coin or PayPal): /sponsor, POST /api/v1/sponsor, peptides sponsor <vendor> --email <you> or the MCP tool sponsor_vendor.

MIT.