TG
Tskgone
NewMCP + OAuth 2.0 support

Distributed Task Scraping Platform

Tskgone orchestrates scraping across executors, routes crawls by task type, and exposes a signed REST API so agents and pipelines push tasks and pull structured data — without reinventing a job queue.

reddit-crawlPUBLIC
42 / 118
Processing · 3
TSK-1042r/dataisbeautiful · post commentsPROCESSING
0:12
TSK-1041r/programming · subreddit listingPROCESSING
0:34
TSK-1039u/spez · user detailsPROCESSING
1:02
Pending · 71
TSK-1044r/MachineLearning · search 'agents'PENDING
—
TSK-1045r/rust · subreddit rulesPENDING
—
Completed · 44
TSK-1038r/webdev · post detailsCOMPLETED
2m
RedditChrome ExtensionMCPOAuth 2.0HMAC-SHA256

A job queue built for scraping

Tskgone ships the primitives every scraping team rebuilds — task groups, executor pools, progress heartbeats, retries, and typed schemas — as one platform, so you ship the crawler, not the infrastructure.

01

Distributed task queue

Push scraping tasks into named groups. Approved executors pull the next PENDING task with row-level SKIP LOCKED guarantees — no duplicates, no double work.

02

Typed task contracts

Every task type ships a JSON Schema and a documented output shape, so a client knows what to send and what comes back before it writes a line of code.

03

Heartbeat timeouts

Executors post progress heartbeats to extend their lease. Tasks that go silent are auto-failed by a background scanner and become retryable.

04

Typed task types

Each task type ships a JSON schema for input and output. Front-ends and agents get autocomplete; bad payloads are rejected before a worker sees them.

05

File uploads & artifacts

Executors upload screenshots and JSON blobs with magic-byte validation. Files are scoped to the project and retrievable via signed requests.

06

OAuth, MCP & API keys

Expose every project, group, and task as MCP tools. Authenticate with OAuth 2.0, MCP, or project-level API keys — whichever your workflow needs.

Signed REST API

Every call signed. Every payload typed.

HMAC-SHA256 on every request, a 5-minute skew window against replay, and JSON-schema validation before a worker ever sees the task. Rotate credentials on demand.

POST /api/v1/tasks
curl -X POST $TSKGONE/api/v1/tasks \
  -H "X-API-Key: pk_live_…" \
  -H "X-Signature: hmac-sha256(…)" \
  -d '{ "group": "reddit-crawl",
        "type": "post_comments",
        "input": { "url": "…" } }'

# → 201 { "id": "TSK-1042",
        "status": "PENDING" }

From zero to a running scraper

1.0

Create a project

Projects hold your task groups, executors, and generated HMAC credentials. Invite teammates as ADMIN or MEMBER.

2.0

Register executors

Workers register with a UUID and declare the task types they handle. Approve once; they pull on their own after that.

3.0

Push and pull tasks

POST tasks to the signed API and read outputs back via GET — or receive a webhook the moment a task finishes.

Frequently asked

What is Tskgone?

Tskgone is a distributed task scraping platform. You define scraping task groups, executors pull PENDING tasks with HMAC-authenticated calls, and Tskgone tracks progress, timeouts, and execution history.

How is it different from a generic job queue?

Tskgone is scraping-aware: typed task types (Reddit posts, comments, subreddit info/rules/listings, user details, search), per-task-type timeout tuning, and a lease/retry model built for browser executors.

How do executors authenticate?

Every request is signed with X-API-Key, X-Timestamp, and X-Signature (HMAC-SHA256 of method + path + timestamp + body). A 5-minute skew window prevents replay.

Does it support OAuth or MCP?

Yes. /api/mcp exposes every project, group, and task tool to MCP clients, authenticated via user API keys or OAuth 2.0 access tokens with per-scope gating.

What task types are available today?

Reddit post details, post comments, subreddit info, rules, listings, user details, and search. More platforms (X/Twitter and beyond) are on the roadmap.

Build your scraping pipeline today