agentleFS
GhostPawJSGhostPawJS/telegram/AGENTS.md

telegram

AGENTS.md1 starsChanged 7 days agoMITView on GitHub

Telegram transport, SQLite message mirror, rendering, and structured tools for Node.js. Install with npm install @ghostpaw/telegram in a consuming project. The caller owns the database connection. For file databases, enable WAL and busy_timeout=5000 before use. The package does not open a file database. Tools live in src/tools/ and are exported through tools. tools.telegramTools contains synchronous plan handlers. Results have { ok, outcome, summary, next? } plus data on success or error on failure. tools.createTelegramToolFacade({ client: bot.client, execute: true }) adds…

  • Reads credentials
  • Installs packages
  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,843 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository2 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use Node 24. npm run check is the single gate: types, lint, tests, build, zero runtime dependencies, generated surfaces, size budgets, API …

    line 49
  2. Very specific:

    Use node:test and node:assert, with *.test.ts next to source for unit tests.

    line 52
  3. Distinctive:

    Keep exported APIs, tool names and fields, CLI flags, database schemas, and synchronous behavior compatible.

    line 59

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

@ghostpaw/telegram

Telegram transport, SQLite message mirror, rendering, and structured tools for Node.js.

Use

Install with npm install @ghostpaw/telegram in a consuming project.

import { DatabaseSync } from 'node:sqlite';
import { createBot, initTelegramTables, write } from '@ghostpaw/telegram';
const db = new DatabaseSync('telegram.db');
db.exec('PRAGMA journal_mode = WAL; PRAGMA busy_timeout = 5000');
initTelegramTables(db);
const bot = createBot({ token: process.env.TELEGRAM_TOKEN!, db });
await write.sendMessage(bot.client, 123, 'Hello');

The caller owns the database connection. For file databases, enable WAL and busy_timeout=5000 before use. The package does not open a file database.

Tools live in src/tools/ and are exported through tools. tools.telegramTools contains synchronous plan handlers. Results have { ok, outcome, summary, next? } plus data on success or error on failure. tools.createTelegramToolFacade({ client: bot.client, execute: true }) adds an async handler(db, name, input) that executes send and moderation plans. Omitting execute keeps plan-only behavior. The facade defaults to one action per second per chat, 30 globally, and two retries for Telegram retry_after responses. Configure perChatPerSecond, globalPerSecond, maxRetries, now, and sleep at construction.

Layout

  • src/bot.ts: bot lifecycle and incoming updates.
  • src/bot_api/, src/telegram_types.ts: fetch client, uploads, and structural Bot API types.
  • src/executor/: outbound actions; src/streaming/: progressive messages.
  • src/messages/, src/chats/, src/users/, src/members/: SQLite storage.
  • src/render/, src/normalize/: rendering and incoming data conversion.
  • src/tools/, src/skills/: structured tool definitions and workflows.
  • tests/: built ESM/CommonJS integration checks (build before testing).
  • src/surfaces/: shared MCP, stdio, CLI, skill and manifest renderers.
  • src/surface/: Telegram bindings; src/mcp.ts, src/cli.ts, src/mcp_server_bin.ts are entries.
  • scripts/surfaces/: generates SKILL.md, skills/telegram/, plugin.json, mcp.json, .claude-plugin/, server.json, glama.json, and mcpb/manifest.json. Run npm run surfaces; do not edit generated outputs.
  • scripts/guard/: size, API, consumer, startup and ratchet guards; scripts/bench/: benchmarks.
  • docs/: detailed usage documentation.

Contribute

Use Node 24. npm run check is the single gate: types, lint, tests, build, zero runtime dependencies, generated surfaces, size budgets, API snapshot, packed consumer, startup budgets, and ratchet. Also run npm run test:dist. Use node:test and node:assert, with *.test.ts next to source for unit tests. Built artifact integration tests live in tests/; run npm run test:dist after building. Assert real behavior; preserve existing assertions and coverage.

The package has zero runtime dependencies. Do not add runtime dependencies. npm run check:deps checks the manifest and built package. grammy and marked are development dependencies for compatibility tests only. Keep exported APIs, tool names and fields, CLI flags, database schemas, and synchronous behavior compatible. Make additive changes only. Budget increases require [budget-increase-approved], API snapshot changes require [api-change-approved], and guard, workflow or lint changes require [guardrails-change-approved] in the reviewed commit message. Record changes under Unreleased in CHANGELOG.md in neutral maintainer prose. Leave .github/workflows/disabled intact.

Release with npm run release:patch or npm run release:minor; these scripts version and tag the package and push so CI publishes with provenance. Maintainers run releases after review; leave task changes uncommitted.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.