GoGogot is live · free · last verified 24 Oct 2026 · viability monitor
GoGogot
GoGogot is an open-source, self-hosted AI agent in Go — 8,835 lines, 8 dependencies, 34 tools, answering you in Telegram.
The readable codebase is the product here. At 8,835 lines and 8 direct dependencies, one person can audit the loop that has a shell on their server — that is the argument, not a feature list. What you give up is real: single-owner Telegram auth, no team mode, no permissions, and recall that hands the model past-chat summaries instead of doing a vector search. Pick it over OpenClaw or Nanobot if you want an agent you can fork; skip it if you need a UI anyone else on your team will use.
Verdict generatedVerified 6d ago · liveness 63/100 · cite: rightaichoice.com/tools/gogogot
- Solo developers who want an automation agent they can audit line by line on their own server
- DevOps engineers who prefer Telegram chat-ops to a web dashboard
- Privacy-minded operators who want model choice, data and keys to stay on their box
- Go developers who intend to read, fork or extend the source — adding a tool is a struct and a function
- Teams needing multi-user access, roles or shared workspaces — TELEGRAM_OWNER_ID is a single integer
- Anyone who needs a graphical web UI or a native mobile app to operate the agent
- Non-technical users unwilling to deal with Docker, env vars and the terminal
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip GoGogot if you need more than one person to operate the agent — TELEGRAM_OWNER_ID is a single integer, and there are no roles, shared workspaces or web UI for teammates to log into.
GoGogot is free, but every session bills your model provider — the vendor's own table puts a DeepSeek V4 Flash session at roughly $0.01 and a claude-opus-4-8 session at roughly $0.50, assuming about 50K tokens in and
GoGogot itself is MIT-licensed and free — no account, no seats, no subscription. The real cost line is your model provider: the vendor's own table runs from roughly $0.01 per session on deepseek/deepseek-v4-flash (the recommended default, 1M context, 81.7% PinchBench) up to roughly $0.50 on claude-opus-4-8, plus whatever you pay for the host — the site benchmarks against a ~$4 VPS. Frontier-model users should expect well under a dollar per session; heavy tool loops cost more.
In short
GoGogot — GoGogot is an open-source, self-hosted AI agent in Go — 8,835 lines, 8 dependencies, 34 tools, answering you in Telegram. Best for Solo developers who want an automation agent they can audit line by line on their own server, DevOps engineers who prefer Telegram chat-ops to a web dashboard, Privacy-minded operators who want model choice, data and keys to stay on their box. Free to use.
What is GoGogot?
GoGogot — GoGogot is an open-source, self-hosted AI agent in Go — 8,835 lines, 8 dependencies, 34 tools, answering you in Telegram. Best for Solo developers who want an automation agent they can audit line by line on their own server, DevOps engineers who prefer Telegram chat-ops to a web dashboard, Privacy-minded operators who want model choice, data and keys to stay on their box. Free to use.
GoGogot is an open-source, self-hosted AI agent written in Go for developers and sysadmins who want their automation to live on their own box instead of someone else's cloud. The whole agent is one for loop in internal/core/agent/run.go: call the model, run whatever tools it asked for, feed the results back, repeat until it stops asking. Vendor counts from the repository on 2026-08-26 put that at 8,835 lines of Go across 82 files, 31 packages, 8 direct dependencies and 34 tools — shell and filesystem (bash, read_file, write_file, edit_file, list_files, file_search, system_info), four web tools (web_search, web_fetch, web_request, web_download), plus memory, skills, identity, scheduling, conversation and planning tools. Memory, scheduling and identity are not subsystems; they are tools the model calls from inside that same loop. Interaction is Telegram, and it opens no inbound ports — no listener, no webhook, no EXPOSE. The agent polls Telegram and calls your model provider, so every connection is outbound; auth is TELEGRAM_OWNER_ID, a single integer, and everyone else who messages the bot is ignored. You pick the brain with three environment variables (GOGOGOT_PROVIDER, GOGOGOT_MODEL and the matching API key) and restart the container. Eight aliases work out of the box, from deepseek/deepseek-v4-flash at roughly $0.01 per session (about 50K tokens in, 10K out) up to claude-opus-4-8 at around $0.50. Everything it remembers lives as plain files under GOGOGOT_DATA_DIR: soul.md, user.md, markdown memory, saved skills, episodes as json+jsonl pairs and schedules.json — there is nothing to export because you already have it. The pitch is scale by subtraction. Against OpenClaw (430K lines, 800+ dependencies) and Nanobot (4K lines, 30+ dependencies), GoGogot claims 8.8K lines, 8 dependencies, a 203 MB compressed image and ~10 MB RAM at idle. The trade is honest: no plugin system, no embeddings behind recall, no multi-user mode, and the reasoning still runs on someone else's model API.
Key features
- Self-hosted AI agent in Go — 8,835 lines across 82 files, 31 packages, 8 direct dependencies
- Core loop in internal/core/agent/run.go: call model, execute tools, feed results back, repeat until the model stops asking
- 34 built-in tools across shell, files, web, memory, skills, identity, scheduling, conversation and planning
- Shell and file tools: bash, read_file, write_file, edit_file, list_files, file_search, system_info
- Four web tools: web_search, web_fetch, web_request, web_download (web_search needs BRAVE_API_KEY)
- Memory tools memory_list, memory_read, memory_write, memory_delete backed by plain markdown files
- Skills system: skill_create, skill_update, skill_read, skill_list, skill_delete — save a procedure once and rerun it by name
- Identity tools soul_read, soul_write, user_read, user_write — the agent rewrites soul.md itself
- Cron scheduling via schedule_add, schedule_list, schedule_pause, schedule_remove, persisted to schedules.json
- Telegram controls: send_message, send_file, ask_user, report_status, recall
- task_plan tool for multi-step work inside a session
- Swap models with GOGOGOT_PROVIDER, GOGOGOT_MODEL and the matching API key, then restart the container
- Eight model aliases out of the box, including qwen3.7-max, claude-opus-4-8, deepseek-v4-flash and gpt-5.4-mini
- No inbound ports — polls Telegram and calls the model provider, so all traffic is outbound
- All state as plain files in GOGOGOT_DATA_DIR: soul.md, user.md, memory/, skills/, episodes/, schedules.json
Behind the verdict
GoGogot's bet is that agent frameworks got big because they tried to be platforms, and that a single-operator agent can be small on purpose. The homepage makes the case with five linear measurements against OpenClaw and Nanobot: 8.8K lines versus 430K and 4K, 8 dependencies versus 800+ and 30+, plus install size, idle RAM and a compressed Docker image reported at 203 MB. That is a marketing frame, but it is a checkable one — the counter on the page says the 8,835-line figure was counted from the repository on 2026-08-26, tests excluded, and invites you to re-count. The technical shape matches the pitch. Agent.Run takes the message list, calls a.llm.Call with the tool definitions, appends whatever tool calls come back, executes them, appends the results and loops until the model stops asking. Memory, identity and cron are registered as tools, not carved out as services with their own storage layers — memory_list, memory_read, memory_write, memory_delete hit plain markdown files; soul_read and soul_write let the agent rewrite soul.md itself; schedule_add and friends persist to schedules.json. The full tool list is 34 names across nine groups: shell and files (bash, read_file, write_file, edit_file, list_files, file_search, system_info), web (web_search, web_fetch, web_request, web_download), memory, skills (skill_create, skill_update, skill_read and friends), identity, scheduling, conversation (send_message, send_file, ask_user, report_status, recall) and planning (task_plan). If you have ever wanted to trace exactly what an agent did with your shell, this is a codebase small enough to do it in an afternoon. Running it is deliberately unglamorous. There is no inbound listener, no webhook and no EXPOSE — the agent polls Telegram and calls your provider, so all traffic is outbound and you never open a port. Auth is a single integer, TELEGRAM_OWNER_ID; anyone else messaging the bot is ignored. Model choice is three env vars and a container restart, and the published table lists eight working aliases with context, per-session cost and a PinchBench agent success score: qwen/qwen3.7-max (1M context, ~$0.10/session, 92.5%), claude-opus-4-8 (1M, ~$0.50, 90.5%), nvidia/nemotron-3-ultra-550b-a55b (1M, ~$0.05, 89.9%), qwen/qwen3.6-flash (1M, ~$0.02, 88.1%), deepseek/deepseek-v4-flash (1M, ~$0.01, 81.7%, the recommended default), google/gemini-3.1-pro-preview (1M, ~$0.22, 81.0%), gpt-5.4-mini (400K, ~$0.08, 79.2%) and gpt-5.4-nano (400K, ~$0.02, 69.0%). The vendor is careful to note PinchBench was measured with a different agent harness, so read the column as a model ranking, not a GoGogot ranking. The honest failure modes are all on the limits section. No web UI and no mobile app — Telegram or the terminal. No multi-user, no roles, no shared workspaces. No built-in RAG, no vector store, no multi-agent orchestration; recall is summaries, not nearest-neighbour search. No plugin system — the vendor's own words are that adding a capability is "a struct and a
Models under the hood
as of 2026-10-30
Limitations
- Self-hosted and single-operator: no web UI or mobile app, interaction is via terminal or Telegram (with Discord, Slack and email named as pluggable interfaces the agent can write plugins for).
- Model spend is your own — the site quotes roughly $0.01 per session and ~10 MB RAM at idle, with heavier tool loops costing more.
- The model does not run locally; prompts go to whichever external provider you configure via GOGOGOT_PROVIDER/GOGOGOT_MODEL.
as of 2026-10-14
Researching GoGogot? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is GoGogot still active?
GoGogot is live · free · last verified 24 Oct 2026 · viability monitor.
Viability score
How well maintained and how widely used is GoGogot? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What do users say about GoGogot?
What people actually say about GoGogot — is it worth it?
Our community sentiment research written Jul 4, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about GoGogot— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 24 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified GoGogot 11 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed limitations, pricing page
- — changed SSL certificate
- — changed setup time, description, our verdict, features, hidden costs, integrations, limitations, switching in, switching out, who should skip it, pricing tiers, pricing fit, summary
- — changed integrations, who it suits, description, summary, who should skip it, features, our verdict, pricing tiers
- — changed who should skip it, limitations, who it suits, switching in, switching out, models, use cases, pricing fit, description, setup time, our verdict, features, summary, hidden costs
- — changed domain expiry
Showing the 6 most recent of 11 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does GoGogot cost in 2026?
GoGogot: Free to use. Pricing last verified 24 Oct 2026.
Plans compared
Every published GoGogot tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Open source (self-hosted) | FreeVendor: “$0 (MIT)” | — | Solo developers, sysadmins and Go programmers who want an auditable agent on their own VPS, Raspberry Pi or home server and are comfortable with Docker, env vars and the terminal.Adds: Starting tier and the only tier: MIT-licensed source you can read and fork, no account and no seats, with model cost billed directly by your provider (roughly $0.01 per session on deepseek-v4-flash). |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where GoGogot's pricing actually pencils out — and where peers do it cheaper.
GoGogot itself is MIT-licensed and free — no account, no seats, no subscription. The real cost line is your model provider: the vendor's own table runs from roughly $0.01 per session on deepseek/deepseek-v4-flash (the recommended default, 1M context, 81.7% PinchBench) up to roughly $0.50 on claude-opus-4-8, plus whatever you pay for the host — the site benchmarks against a ~$4 VPS. Frontier-model users should expect well under a dollar per session; heavy tool loops cost more.
Has GoGogot's pricing changed?
No plan price for GoGogot has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick GoGogot?
Pick GoGogot if you are Solo developers who want an automation agent they can audit line by line on their own server, DevOps engineers who prefer Telegram chat-ops to a web dashboard, Privacy-minded operators who want model choice, data and keys to stay on their box, Go developers who intend to read, fork or extend the source — adding a tool is a struct and a function.
Look elsewhere if you are Teams needing multi-user access, roles or shared workspaces — TELEGRAM_OWNER_ID is a single integer, Anyone who needs a graphical web UI or a native mobile app to operate the agent, Non-technical users unwilling to deal with Docker, env vars and the terminal, Knowledge-base workloads that need embeddings or a vector store — recall is not a vector search.
Skip it if Skip GoGogot if you need more than one person to operate the agent — TELEGRAM_OWNER_ID is a single integer, and there are no roles, shared workspaces or web UI for teammates to log into.
GeneratedReal-world workflow fit
Concrete scenarios for the personas GoGogot actually fits — and what changes day-one when you adopt it.
You send one Telegram line: grab the sales CSV from this link, total it by region, send me a PDF. The loop calls web_download to fetch it, bash head -3 to inspect the columns, write_file to generate a pandas script, bash to run it, then send_file to deliver the PDF to chat.
Outcome: Six iterations, five tool calls, about 3.5 seconds, and a four-page regional PDF in the chat — with the agent adding its own read on which region moved.
You use schedule_add to register an hourly task: check disk and memory, ping me above 80%. The cron entry lives in schedules.json and fires through the same loop the model uses for interactive turns.
Outcome: Threshold alerts arrive in Telegram without you opening a monitoring UI, and the schedule is a file you can read or pause with schedule_pause.
You clone the repo and read internal/core/agent/run.go — the call-model / execute-tools / append-results loop is short enough to follow in one sitting — then add your own tool as a struct and a function.
Outcome: You end up with a running agent on your own box and a working understanding of tool-calling loops you can reuse in your own projects.
Use cases
- Automate routine ops from Telegram: hourly disk and memory checks that ping you above 80%
- Turn a CSV link into a finished regional PDF report without leaving chat
- Merge a batch of screenshots into a single PDF and send it back
- Post a daily 9am digest of the five biggest AI stories, two sentences each
- Scrape every email and phone number off a page into a CSV
- Run a Friday 6pm commit summary and changelog from the repo
- Run a small agent loop on a Raspberry Pi or cheap VPS without a cloud account
- Read and fork the entire agent codebase to learn how a tool-calling loop is wired
Setup time & first value
How long it actually takes to get something useful out of GoGogot — broken out by persona, not the marketing-page minute.
Solo developer: minutes to a working agent if you have Docker and a provider key — set GOGOGOT_PROVIDER, GOGOGOT_MODEL and the matching API key, mount GOGOGOT_DATA_DIR, start the container and message the bot from the TELEGRAM_OWNER_ID account. Expect longer if you are provisioning the host or a Brave API key for web_search. Non-technical operators should budget considerably more: there is no web
Switching to or from GoGogot
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From a cloud agent subscription: run the Docker image on your own VPS, point GOGOGOT_PROVIDER and GOGOGOT_MODEL at your provider, and move the tasks you care about into schedule_add entries in schedules.json.
- →From OpenClaw: the tool surface is smaller by design — 34 tools, no plugin system — so port the handful of workflows you actually use and expect to write Go for anything bespoke.
- →From Nanobot: the loop shape is comparable, but GoGogot ships 8 direct dependencies and a published image with python3 and pandas included, so shell-heavy tasks move over with less packaging work.
- →From a hand-rolled cron plus shell scripts: wrap each script as a skill (skill_create) or a schedule_add entry so the model can invoke it by name from Telegram.
- →From a shared team agent: expect to give up multi-user access — TELEGRAM_OWNER_ID is a single integer, so pick one operator account per instance.
- ↗To OpenClaw: if you outgrow 34 tools and want a plugin ecosystem, you are trading a readable 8.8K-line codebase for 430K lines and 800+ dependencies.
- ↗To a hosted agent platform: if you need a dashboard teammates can log into, you are trading self-hosting and BYO-model for someone else's UI and pricing.
- ↗To a RAG stack: if recall-by-summary stops being enough, you need a vector store and embeddings alongside or instead of GoGogot.
- ↗To your own Go service: the loop is short enough to reimplement, and all state is plain files in GOGOGOT_DATA_DIR, so there is nothing to export.
Integrations
Tutorials & Learning
YouTube returned 6 videos for “GoGogot”, and we withheld 6: 6 could not be judged, because “GoGogot” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about GoGogot.
Official links
What are the best GoGogot alternatives?
The closest alternatives to GoGogot we list are Nanobot, Hermes-Agent and Hermes Desktop. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Nanobot | Free | Free | — |
| Hermes-Agent | Freemium | Pricing not published | — |
| Hermes Desktop | Free | Free | — |
Alternatives to GoGogot
View all
Nanobot
Nanobot is an MIT-licensed open-source AI agent you self-host and run in your terminal, browser WebUI, or your own app.
Hermes-Agent
Hermes Agent: a self-hosted open-source AI agent with persistent memory that builds reusable skills across 20+ chat platforms.
Hermes Desktop
Hermes Desktop is a free, open-source desktop GUI for the Hermes Agent — a self-hosted AI agent with a built-in learning loop that turns
Tools that pair well with GoGogot
Common stack mates teams adopt alongside GoGogot, with the specific reason each pairing earns its keep.
Nanobot
Nanobot is an MIT-licensed open-source AI agent you self-host and run in your terminal, browser WebUI, or your own app.
Hermes-Agent
Hermes Agent: a self-hosted open-source AI agent with persistent memory that builds reusable skills across 20+ chat platforms.
Hermes Desktop
Hermes Desktop is a free, open-source desktop GUI for the Hermes Agent — a self-hosted AI agent with a built-in learning loop that turns
GoGogot vs its closest rival
GoGogot's closest head-to-head is with Spider Cloud — Pick Spider Cloud if you need a robust web scraping API for AI data pipelines — it offers structured output, a large scraper catalog, and data connector integrations. Choose GoGogot if you want a lightweight, self-hosted agent for task automation with no cloud dependency and full privacy control. They solve different problems: Spider Cloud fetches external web data; GoGogot executes autonomous actions locally. Verdict generated
Head-to-head comparisons
Gogogot vs Spider Cloud
Pick Spider Cloud if you need a robust web scraping API for AI data pipelines — it offers structured output, a large scraper catalog, and data connector integrations. Choose GoGogot if you want a lightweight, self-hosted agent for task automation with no cloud dependency and full privacy control. They solve different problems: Spider Cloud fetches external web data; GoGogot executes autonomous actions locally.
GeneratedGogogot vs Temporal Ai
Temporal AI is a full-featured durable execution platform for teams that need reliability across complex AI agent workflows and microservices, with strong SDK support and integrations. GoGogot is a minimal, self-hosted alternative for developers who prioritize privacy and lightweight automation without cloud dependencies. Choose Temporal if you need enterprise-grade resilience and collaboration; choose GoGogot if you want a free, Go-native CLI agent for local or edge use.
GeneratedGogogot vs Presto Voice
If you run a QSR chain and want to boost drive-thru revenue with proven voice AI (including a recent Dairy Queen partnership), Presto Voice is the enterprise choice. If you're a developer needing a lightweight, privacy-first task automation agent you can self-host, GoGogot is a free open-source option. These tools serve entirely different markets.
GeneratedFrequently asked questions
Best-of guides
Used GoGogot? Help shape our editorial sentiment research.