Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add genvjacobc/lightspeed-x-mcp --skill writesgit clone --depth 1 https://github.com/genvjacobc/lightspeed-x-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/genvjacobc/lightspeed-x-mcp/writes)<a href="https://agentmods.dev/skills/genvjacobc/lightspeed-x-mcp/writes"><img src="https://agentmods.dev/badge/skills/genvjacobc/lightspeed-x-mcp/writes/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/genvjacobc/lightspeed-x-mcp/writes"><img src="https://agentmods.dev/badge/skills/genvjacobc/lightspeed-x-mcp/writes.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00124 | $0.01317 |
| Opus 5.5 | $0.00050 | $0.00527 |
| Sonnet 5.5 | $0.00025 | $0.00263 |
| Haiku 4.5 | $0.00012 | $0.00132 |
Grade A, and why
writes scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 18d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 88 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Changing data in Lightspeed X
Every write here is two steps, and only the second one changes anything.
- Preview. Call the purpose-built write tool. It reads current state,
validates, and returns a diff, the exact requests, a
plan_idand aconfirmphrase. Nothing is sent. - Apply. After the user has seen the preview and said yes, call
lightspeed_apply_writewith thatplan_idand theconfirmphrase copied exactly. The user is asked to approve that call every time.
Rules
- Never apply a plan the user has not seen. Show the preview's summary, warnings and request list in your reply, then stop and wait for an explicit yes. "Go ahead and fix it" before a preview exists is permission to preview, not to apply.
- One plan, one yes. If the user changes anything about the request, run the preview again; do not apply an older plan.
- Copy the confirm phrase exactly. It is written by the server and shown in the approval prompt so the user can see what they are approving. Never paraphrase it or build one yourself.
- Relay warnings and UNVERIFIED labels. If the preview says an endpoint is unverified, or that a step cannot be undone, say so plainly before asking for a yes. After an unverified write, tell the user to check the back office.
- If the user declines, stop. Do not retry with a different tool, the raw write tool, or a smaller batch to get around a no.
- Prefer the specific tool over
lightspeed_api_write. The raw tool validates nothing, records no undo, and should be a last resort the user understands.
Pick the tool
| They want to... | Tool |
|---|---|
| Add/remove tags on products | lightspeed_update_product_tags (merges by name, creates missing tags) |
| Change a product's name, price, brand, category, active flag | lightspeed_update_product |
| Create a product or variant family | lightspeed_create_product |
| Stop selling something | lightspeed_update_product with is_active: false (prefer over delete) |
| Create or edit a promotion | lightspeed_upsert_promotion |
| Retire a promotion | lightspeed_archive_promotion (there is no promotion delete) |
| Brands, suppliers, tags, categories, price books, webhooks | lightspeed_manage_* |
| Move stock between stores | lightspeed_create_consignment type OUTLET, then lightspeed_advance_consignment SENT, then RECEIVED at the destination |
| Raise a purchase order | lightspeed_create_consignment type SUPPLIER |
| Receive a delivery | lightspeed_advance_consignment RECEIVED with received counts, or receive_all |
| Write off damage/theft, add found stock | lightspeed_adjust_stock (quantity is a delta, sign must match the reason) |
| Recount a store | lightspeed_create_consignment type STOCKTAKE |
| Reorder rules | lightspeed_set_reorder_points |
| Customers and groups | lightspeed_upsert_customer, lightspeed_customer_group_members |
| Store credit, gift cards, loyalty, returns, voids, registers | high-risk tools; see below |
| Undo something applied earlier | lightspeed_revert_write with the old plan_id, then apply |
| See what is enabled, pending, or recently changed | lightspeed_write_status |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 18d ago First seen · 88 lines · 124 tokens per session scan A 7cf2a8986ee0
writes is a skill published in the GitHub repository genvjacobc/lightspeed-x-mcp (0 stars, last pushed 19d ago), licensed MIT. It adds 124 tokens to every session and 1,317 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-09.
Other skills, from other repositories
warehouse-management
A warehouse-management workflow for МойСклад, an online system for products, stock, orders, counterparties, and financial documents. It can search records, inspect stock and reports, and create operational documents.
recurrent-payment
A recurring-payment workflow for T-Kassa, a payment service. It links a customer's card during an initial payment and later charges that saved payment authorisation for a subscription.
wb-feedbacks
A Wildberries customer-feedback workflow for reviews, questions, and return requests. Wildberries is an online marketplace; the workflow groups feedback, prepares replies, and connects return reasons with reported product problems.
inbox-triage
A procedure for reviewing unread Avito chats and connecting each conversation to its listing. Avito is an online marketplace where people buy and sell items.
my-listings
An Avito listing-review workflow. Avito is an online classifieds marketplace, and this lists your advertisements by status and category.
card-payment
A Payme card-payment workflow for UzCard and Humo cards. It saves a card token, confirms the card by SMS, creates a receipt, and charges the card.