Connect Hermes Agent to LinkedIn - safely (2026)
Yes, and without the ban risk. Zevari gives your Hermes Agent LinkedIn over MCP or a REST API - search, score, draft, and run campaigns - with every send staged for your approval by default. No browser cookies. Published limits. A year in production, zero ban incidents.
This is about Hermes Agent, the autonomous agent framework from Nous Research - not the separately named open-weight LLM family.
Give Hermes Agent the tools (config.yaml)
# ~/.hermes/config.yaml (/opt/data/config.yaml in Docker)
mcp_servers:
zevari:
url: "https://mcp.zevari.ai/mcp"
headers:
Authorization: "Bearer ${ZEVARI_MCP_TOKEN}"Or call REST from the agent
$ curl https://api.zevari.ai/v1/linkedin/searchPosts \
-H "Authorization: Bearer $ZEVARI_API_KEY" \
-d '{"keywords":"hiring an SDR","date_posted":"past_month"}'Will this get my account banned?
It is the first question, and it is the right one. Most "Hermes Agent for LinkedIn" setups earn the ban because they point the agent at a logged-in browser session with your cookies - the exact mechanism LinkedIn flags. 2026 was a culling: HeyReach was cut off by LinkedIn in March, Apollo and Seamless before it, detection is up 340% since 2023, and in one 50-account test 23% were restricted inside 90 days.
Zevari was built to be the safe one, and after a year of refinement it has zero ban incidents. Safety here is receipts, not adjectives:
- By default every write - message, connection request, comment, post - is staged for your approval, even from a self-improving Hermes Agent. A specialist moves to autopilot only after it has earned it, and you can switch it back at any time.
- Session-based connection. No stored browser cookies, ever. No password handoff, no extension.
- Weekly connection ceilings, enforced server-side: 150 on Free* and Premium, 200 on Sales Navigator.
- * Free accounts: LinkedIn caps connection requests that carry a note at 5 a month. The weekly ceiling covers requests sent without a note, and Premium and Sales Navigator accounts get a far larger allowance for requests with a note. Most outreach uses a note, so Premium or Sales Navigator is the practical choice for real outbound - the full explanation is on /safety.
- Working hours and behavioral pacing, plus duplicate checks and burst caps.
Setup, about 10 minutes
How to connect Hermes Agent to LinkedIn in four steps
- 1
Create a Zevari MCP access token
Sign in to Zevari and open Settings → Developers. On the MCP access tokens tab, create a token for this agent. It starts with zvr_mcp_ and is shown once, so store it in the agent's environment as ZEVARI_MCP_TOKEN, for example in ~/.hermes/.env. The Zevari MCP server accepts it as a bearer token in the Authorization header, so an agent with no browser never needs the sign-in flow - no LinkedIn password, no browser cookies, no extension. For the REST path, create a workspace key on the API keys tab and store it as ZEVARI_API_KEY.
- 2
Register Zevari in your Hermes Agent config (MCP or REST)
Hermes Agent supports MCP servers natively, so add Zevari to the mcp_servers block in ~/.hermes/config.yaml (/opt/data/config.yaml in Docker), with the MCP access token in an Authorization header, and the agent discovers all 191 LinkedIn and GTM tools automatically. Or call the REST API directly with your API key from the agent's own code. The MCP path is the cleanest; the REST path is the universal one for any runtime.
Both paths, with copy-paste config, are in Two ways in below.
- 3
Verify the connection
Confirm the agent can see Zevari before you run anything live. A read-only call - get the safety status, or search for recent signal posts - proves the token works and returns your workspace context, Voice DNA, and the enforced weekly ceilings.
terminal$ curl https://api.zevari.ai/v1/safety/getStatus \ -H "Authorization: Bearer $ZEVARI_API_KEY" # -> your connection state, weekly ceiling for your account tier, and open incidents - 4
Run your first safe loop: find, score, stage, approve
Tell the agent what you want: find people who posted about a topic in the last 30 days, ICP-score them, draft in your voice, and stage a campaign. By default every send is held at the approval gate, even with a persistent, self-improving loop - you approve from chat, Slack, or a daily digest, and the agent executes inside your ceilings and working hours. A specialist sends on its own only after you move it to autopilot once it has earned it, such as Follow-up after ten approved sends in 30 days, and you can switch it back at any time.
Connect Hermes Agent to LinkedIn
Two ways in
Zevari is hosted - nothing to self-host, no scraper to babysit, no cookie to refresh. Hermes Agent supports MCP natively, so add it to your config.yaml with an MCP access token, or call the REST API with an API key.
Add the Zevari server with your MCP access token as the bearer header, and Hermes Agent discovers 191 LinkedIn and GTM tools automatically:
# ~/.hermes/config.yaml (/opt/data/config.yaml in Docker)
mcp_servers:
zevari:
url: "https://mcp.zevari.ai/mcp"
headers:
Authorization: "Bearer ${ZEVARI_MCP_TOKEN}"Call the same engine with one bearer token from the agent's own code:
$ curl https://api.zevari.ai/v1/linkedin/searchPosts \
-H "Authorization: Bearer $ZEVARI_API_KEY" \
-d '{"keywords":"hiring an SDR","date_posted":"past_month"}'Full endpoint reference in the developer docs. For the bigger picture, see the LinkedIn for AI agents pillar and the LinkedIn MCP page.
A real loop
Find -> score -> stage -> approve
Why approval-gating matters for an autonomous Hermes Agent
Hermes Agent runs a self-improving loop with persistent memory, around the clock. Pointed at LinkedIn unattended, that is a banned account waiting to happen - and a brand risk you can't take back. Most tools in this layer wave it away: ConnectSafely advertises "no approval process required" as if that were the feature. It's the bug, not the feature.
Zevari inverts it. Your agent does all the work - finds, scores, drafts, sequences - and then, by default, stops at the gate. Every write surfaces as a one-tap approval in chat, in Slack, or in a daily digest. You approve in minutes; the agent executes inside your ceilings and working hours. A specialist moves to autopilot only after it has earned it, such as Follow-up after ten approved sends in 30 days, and you can switch it back to reviews with you at any time. Autonomy on the research, human judgment on the send.
FAQ
Hermes Agent and LinkedIn, answered
Will connecting Hermes Agent to LinkedIn get my account banned?
It can if you do it the common way - pointing the agent at a logged-in browser session with your cookies. That is the mechanism behind the 2026 bans: HeyReach was cut off by LinkedIn in March 2026, Apollo and Seamless before it, detection is up 340% since 2023, and in one 50-account test 23% were restricted inside 90 days. Zevari is the other path. Your Hermes Agent reaches LinkedIn through a hosted, session-based connection with no browser cookies, every write is staged for your approval by default, and weekly connection ceilings, working hours, and burst caps are enforced server-side. After a year of refinement it has zero ban incidents. The full mechanics are at /safety.
How do I connect Hermes Agent to LinkedIn - MCP or REST?
Both work. Hermes Agent supports MCP servers natively, so add Zevari to the mcp_servers block in your ~/.hermes/config.yaml (/opt/data/config.yaml in Docker) with your Zevari MCP access token as a bearer Authorization header, and the agent discovers 191 LinkedIn and GTM tools automatically. Or call the REST API directly with an API key from the agent's own code. The MCP path is the cleanest for Hermes Agent; the REST path is the universal one. Same capabilities and the same approval gate on both. The developer docs have the full reference.
Is Hermes Agent the same as the Hermes LLM?
No. Hermes Agent is the open-source autonomous agent framework from Nous Research - persistent memory, a self-improving loop, VPS-hosted, configured at ~/.hermes/config.yaml. The Hermes LLM family is a separate set of open-weight language models. This page is about the agent framework. Zevari gives that framework safe, approval-gated LinkedIn access; the underlying model it points at does not change how the safety works.
My Hermes Agent runs on a VPS. Does that work?
Yes - that is exactly what this is for. Your agent keeps running on your VPS and calls Zevari's hosted endpoint to do the LinkedIn work safely. Zevari holds the campaign schedule and the approval queue on our infrastructure, so sequences advance and the inbox gets classified even between the agent's runs. The agent is yours; Zevari is the LinkedIn layer and persistent scheduler underneath it.
Does the agent actually send, or just draft?
By default it stages. Every message, connection request, comment, and post is queued for your one-tap approval - in chat, Slack, or a daily digest - before anything sends. A fully autonomous Hermes Agent does all the finding, scoring, and drafting, then stops at the gate. A specialist sends on its own only after you move it to autopilot once it has earned it, such as Follow-up after ten approved sends in 30 days, and you can switch it back at any time.
Do I need LinkedIn Sales Navigator?
No. You can start on a free LinkedIn account. Sales Navigator only raises your weekly connection ceiling (150 on Free* and Premium, 200 on Sales Navigator) - it is not required to connect Hermes Agent or run campaigns.
Give Hermes Agent LinkedIn - safely
Run it as code, or you connect your agent. Same approval-gated engine either way.
Connect your agent
You run Hermes Agent. Get the hosted LinkedIn layer over MCP or REST, 191 MCP tools, 109 REST endpoints, and the full safety model, self-serve and live in 60 seconds.
Connect your agentLet the team run it
Don't want to operate an agent? The same six specialists, Prospecting, Signals, ICP Scoring, Voice Writer, Follow-up, and Reply Qualifier, run your LinkedIn and email outbound from the Zevari app. Sends wait for your approval by default.
Build my AI team