which Telegram API each surface speaks, and what each one costs
telegram-dev
Telegram, split by the API each surface actually speaks, in three skills. telegram-bots covers the official HTTP Bot API: update_id as the only idempotency key an update carries, the allowed_updates default that subscribes you to everything except chat_member and the two reaction types while returning ok, the webhook secret header that is the real check, rate limits as a design constraint rather than an error path, and Telegram Stars with a ten-second pre-checkout window and the grant that belongs to successful_payment. telegram-userbots covers MTProto and Telethon: the session file as a credential equal to the password, FloodWaitError as the API telling you exactly how long to wait, pinning across minor releases that move session and entity-cache behaviour, and the account-ban risk a bot token does not carry — opening with whether a user account is needed at all. telegram-miniapps covers the web layer, whose entire security model is one signed query string: verifying initData server-side, the auth_date window, the Ed25519 path for a third party, and the SDK package whose name moved. Both the verifier and the delivery invariants ship as runnable fixtures with their mutants.
What it ships
Each name below is an entry point an agent can be routed to.
- telegram-bots
- telegram-userbots
- telegram-miniapps
Shape: static. Three fixed surfaces with fixed seams; the only branch is which of the three a task is on, and the first table in each SKILL.md decides that before any code.
Install just this one
Every pack installs standalone. The whole family is one command.
npx skills add ssheleg/telegram-dev
claude plugin marketplace add ssheleg/telegram-dev && claude plugin install telegram-dev@telegram-dev
When the agent reaches for it
These are the rules the family writes into your agent's own instruction file — verbatim. Each one states the rule, the boundary in both directions, and the phrase that declines it.
/telegram-dev — which Telegram API a surface speaks, and what it costs
- when: the thing being built lives inside Telegram
- decline it: «без телеграма».
If telegram-dev is installed, work on a Telegram surface goes through it — a bot on the official Bot API, a user account over MTProto, or a Mini App inside the client. It owns the three seams that look alike and are not: update_id is the only idempotency key an update carries, so a redelivery is indistinguishable from a new event; a session file is a logged-in person rather than a token, revocable by the human and bannable by Telegram; and a Mini App's entire authentication is one signed query string that the client hands you already parsed, in a field named initDataUnsafe.
The boundary — "Telegram is the platform, not the transport." NOT through it: sending yourself an alert through a bot, which is one HTTP call; what the bot should do, which is super-ux; how its Mini App looks, which is sheleg-design; charging a card, which is sheleg-dev. Routing every sentence containing the word Telegram through it is how you teach an agent to skip it where the money and the account actually are.
Refusal phrase: «без телеграма».
Among the routers: sheleg-dev wires the money that reaches a card; telegram-dev owns Telegram's own rails — Stars, the bot token, the account — and the risk that sits beside them.
The rest of the family
super-ux v0.48.0
Scenario-driven UI development: a versioned design chain in docs/ux/ — the product vision, personas and jobs, user flows and the paid-acquisition funnel among them, a screens-and-states map with Figma frames…
task-pipeline v1.77.0
Full-cycle delivery orchestrator: an intake grill turns the request into a complete brief, then ten gated stages carry it from docs to acceptance, refusing to advance until each gate passes.
agent-sync v1.17.0
Coordination plane for concurrent agents: leases with a TTL so two agents cannot claim the same work, race-free id reservation, a run journal and a generated board, over a pluggable knowledge cloud.
make-skill v0.23.1
A skill that builds skills: create, retrofit, audit and publish agent skills and Claude Code plugins — Agent Skills standard conformance, marketplace layout, version sync, validator + CI, every distribution…
sheleg-design v1.49.1
The taste layer: cinematic scroll-driven landing pages (one scroll clock, motion that degrades to calm, WebGL particle formations), product-UI style packs with a ready token layer each, and the Figma border…
seo-aeo-audit v0.25.1
Evidence-first website audit for search and answer engines: ten tracks from crawl access to AI citation mechanics, every finding backed by an observation and every recommendation tiered, ending in a…
sheleg-dev v0.10.0
The integration layer a product reaches once it has users: Stripe subscription billing reconciled into your own database — checkout, renewals, seats and proration, refunds, claim-first webhook idempotency…
agent-stack v0.13.2
Production patterns for AI agent systems, in four skills. The orchestrator: tool-calling loops that survive context pressure, pipelines with human checkpoints and resume, provider routing with fallback and…