Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/t3ratech/mcp-session-bridgenpx agentmods add skills/t3ratech/mcp-session-bridge/publish-mcp-serverWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/t3ratech/mcp-session-bridge/publish-mcp-server)<a href="https://agentmods.dev/skills/t3ratech/mcp-session-bridge/publish-mcp-server"><img src="https://agentmods.dev/badge/skills/t3ratech/mcp-session-bridge/publish-mcp-server/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/t3ratech/mcp-session-bridge/publish-mcp-server"><img src="https://agentmods.dev/badge/skills/t3ratech/mcp-session-bridge/publish-mcp-server.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00097 | $0.01576 |
| Opus 5.5 | $0.00039 | $0.00630 |
| Sonnet 5.5 | $0.00019 | $0.00315 |
| Haiku 4.5 | $0.00010 | $0.00158 |
Grade A, and why
publish-mcp-server scanned grade A with 0 findings 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 1mo ago.
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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Listing an MCP server on the public directories
Each directory has its own form, its own limits, and at least one control that is easy to miss. Everything below was learned by submitting to it, including the mistakes.
Use the session_* tools from the T3rnel Session Bridge, in the browser the operator
is already signed into. Most of these directories need an account, and several give no
visible error without one.
The rule that costs the most time
A filled form is not a submission. After clicking submit, check what actually happened, in this order:
- Re-snapshot. If the form's fields are gone, it submitted. If they are still there holding your values, it did not.
- Look for
problemon any field.session_snapshotreportsinvalid,problem,maxLengthandrequired.problemcarries the page's own message — "Description must be at most 400 characters" — fromaria-errormessage,aria-describedby, or the browser's validation message. - Check every control, not just the text inputs. A required
<select>shows asrole: combobox, tag: select. Filtering a snapshot toinputandtextareahides it, and the form then fails for a reason that was on screen the whole time. - If it is still unclear,
session_screenshotand read the page. Do not infer. Twice this was concluded to be "probably needs a login" when the answer was a character limit and an unselected dropdown.
Never tick a paid option. Several of these forms offer expedited review for money, and
at least one renders it beside the free submit. A checkbox's value is "on" whether or
not it is ticked — read checked, which the snapshot reports, and leave paid tiers alone
unless the operator has said otherwise in this session.
Smithery — https://smithery.ai
The web form at smithery.ai/servers/new takes an HTTPS URL only. A stdio server has
no such URL, so that form cannot list one. Publish the bundle instead:
node scripts/build-mcpb.mjs # a zip with manifest.json at its root
smithery mcp publish ./build/<name>-<v>.mcpb -n <namespace>/<slug>
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.
- 1mo ago First seen · 117 lines · 97 tokens per session scan A 154d199196bc
publish-mcp-server is a skill published in the GitHub repository t3ratech/mcp-session-bridge (2 stars, last pushed 27d ago), licensed Apache-2.0. It adds 97 tokens to every session and 1,576 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
browser-automation
Control the user's real browser via Browser Controller. Use when asked to interact with web pages, test UIs, fill forms, or read page content.
chrome-bridge
Recipes for the user's real, logged-in Chrome through the chrome-bridge MCP tools and the zero-token chrome-bridge CLI. Use when a browser task needs a multi-step recipe — a form "until the email arrives", a checkout with a test card, a pixel/GTM check, a cookie-banner audit, redirects after a migration, a page…
melaya-setup
Use for anything involving Melaya. Connecting for the first time; setting up phone or browser control; pairing a phone, installing the app, starting the local runner; driving an Android phone or a browser; building, validating, scheduling or debugging agent pipelines; reading connected services; diagnosing a failed…
paint-artist
Paint or draw a picture in Microsoft Paint (Windows 11) through Melaya Computer Control in ONE melayacomputerbatch call driven by a stored plan. Use when asked to paint or draw something in Paint, to reproduce an image or a painting stroke by stroke, to paint a subject in the style of a painter or of a reference…
helmdeck-hyperframes-authoring
Author render-deterministic hyperframes compositions. Use when generating HTML/CSS/JS for hyperframes.compose, hyperframes.render, or any pipeline that produces a programmatic MP4 via hyperframes — including builtin.scaffolded-narrated-video, builtin.prompt-video, builtin.prompt-narrated-video. The framework's "render…
website-explorer
Discover website capabilities from user behaviors. Learn APIs and automate workflows.