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 agentmods add agents/openagents-org/openagents/openworkergit clone --depth 1 https://github.com/openagents-org/openagentsWrote this? Show the measurements
A badge for your README with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them.
[](https://agentmods.dev/agents/openagents-org/openagents/openworker)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 | $0.00000 | $0.02657 |
| Opus 5 | $0.00000 | $0.01328 |
| Sonnet 5 | $0.00000 | $0.00531 |
| Haiku 4.5 | $0.00000 | $0.00266 |
Grade A, and why
openworker scanned grade A with 1 finding 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 yesterday.
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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
tool the agent does not have is how a model ends up printing curl commands as How it starts
The opening of the file, as written. The whole thing — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.
OpenWorker
OpenWorker (source) run as an OpenAgents agent.
OpenWorker is the odd one out in this catalog. Every other agent here is a
terminal program with a headless mode — claude -p, dsh --profile headless,
command-code -p --output-format json. OpenWorker has no such mode. Its
openworker command is a Textual TUI that
exits with the terminal, and its GUI is a Tauri shell. The only programmable
surface is openworker-server: a FastAPI process that speaks REST for state
and one WebSocket per session for turns.
So OpenAgents runs the server itself. Nothing about that is visible in the workspace — you add the agent, give it a model key, and it answers messages — but it is the fact that explains every design decision below.
What the adapter does
openworker-server --host 127.0.0.1 --port <free> --cwd <working dir> --model <provider:model> --mode bypass-approvals
preflight()runs before the daemon joins the workspace and resolvesopenworker-server. No server, no join — with the install command instead of a poll loop that cannot succeed.- On the first message the adapter provisions a private state directory, mints a launch token, picks a free loopback port, and starts one server. Every channel of that agent shares it; booting the server imports FastAPI, uvicorn and three model SDKs, which is not a cost worth paying per message.
- For each message it opens
ws://127.0.0.1:<port>/ws/session/<id>, sends oneuser_messageframe, and reads events untilturn_done. - The socket closes after the turn. Session state lives on the server, keyed by an id derived from the channel, so a channel keeps its history across daemon restarts without idle sockets being held open.
Authentication of the local socket
The server gates both surfaces on a launch token. OpenAgents supplies its own
through COWORKER_API_TOKEN rather than letting the server write a
sidecar-<port>.token file, so there is no file to race with, leak or clean up.
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.
- yesterday First seen · 198 lines · 0 tokens per session scan A 16ebd6b543b9
openworker is an agent published in the GitHub repository openagents-org/openagents (4,029 stars, last pushed 2d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 2,657 tokens. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-02.
Other agents, from other repositories
worker
General-purpose subagent with full capabilities, isolated context.
deep-review-claude
Independent deep technical review using Anthropic Claude model.
external.anthropic
You are an External Integration agent - a specialized assistant for executing actions across connected external services. Your purpose is to interact with any connected Composio toolkit (900+ platform-wide) and return structured results to the main agent.
external.openai
You are an External Integration agent - a specialized assistant for executing actions across connected external services. Your purpose is to interact with any connected Composio toolkit (900+ platform-wide) and return structured results to the main agent.
explorer.openai
You are an Explorer agent - a specialized workspace investigation assistant. Your purpose is to efficiently gather context from the workspace and return structured findings to the main agent.
explorer.anthropic
You are an Explorer agent - a specialized workspace investigation assistant. Your purpose is to efficiently gather context from the workspace and return structured findings to the main agent.