devglow MCP server
Start, stop, and inspect local dev servers in a desktop app instead of the agent's session.
0 stars152 downloads/wk
Reviews
Write oneNobody has reviewed devglow yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
devglow tools (17, 8 write)
write = sends, deletes, buys or postsRead from the package source without running it. The installed server may list more.
check_portCheck if a TCP port is in use on localhost. When occupied, also reports who holds it — and if it's a devglow-managed project (so the user can recognize their own process). Call this BEFORE run_process / create_project to avoid silent conflicts.
create_projectwrite actionRegister a permanent DevGlow project WITHOUT running it. This is how you 'add a project to DevGlow' — there is no config file to author (no devglow.json / DEVGLOW.md); calling this tool IS the registration. Use it for bulk registration when you want to record metadata only (no port binding, no browser tab, no dependency install). To start the project later, call start_project. Use run_process inst
delete_processwrite actionStop and remove one or more AI processes (the temporary ones created by run_process / workspace_run) — the same as [dismiss] in the DevGlow UI. Use it to clean up processes you started once they are no longer needed. Does NOT apply to permanent projects; use delete_project for those. A pane that was showing a deleted process is left empty.
delete_projectwrite actionDelete a kept (permanent) DevGlow project. Stops the process if running. Does NOT apply to AI processes — use delete_process for those.
delete_workspacewrite actionDelete one or more DevGlow workspaces by id. Removes the arrangement only: the processes that were in it keep running and stay in the sidebar (use delete_process / stop_project for those). The workspace the user is currently in is refused — deleting it would replace their screen. Delete only workspaces you created or the user asked you to remove; list_workspaces first.
get_logsRecent output of a project or AI process, oldest first, one line per entry as [timestamp] [level] message. Works whether it is running or stopped, but DevGlow keeps logs in memory only, so they are gone after the app restarts. Use it when a process exited, failed to start, or behaves unexpectedly. DevGlow writes the logs out on request, so the call can take a few seconds; if they do not arrive, De
get_projectGet the full metadata for a single project or AI process (path, command, port, tags, description, status). Use this when you need details beyond what list_projects shows — for example, to duplicate a project with the same path, to reuse its command in a git worktree (then call run_process with the worktree path), or to read its description before editing.
get_run_historywrite actionRecent finished runs of a project or AI process, newest first, with a summary: how long the command usually takes (median of successful runs), how often it succeeds, and how long it takes to start listening on its port. Use it when the user asks how long a build or script usually takes, whether it has been failing, or what happened on recent runs. Only finished runs are recorded; use get_status fo
get_statusWhether one project or AI process is running right now, with its PID. Lighter than list_projects when you already have the id, for example to confirm that start_project or stop_project took effect (both can return before the change has landed). 'No status found' means DevGlow has no entry with that id. For past runs, use get_run_history.
list_projectsList every registered project and AI process with its running state, port, PID, command, path and id. Projects are the user's permanent entries, started with start_project; AI processes are temporary ones created by run_process or workspace_run. Call this first: it gives you the ids the other tools take, and shows whether the user already has a project for the checkout you are working in. Also lis
list_workspacesList DevGlow workspaces: id, name, which one the user is in, and the processes placed in each with their running state. Use it before delete_workspace, and to find workspaces left over from earlier work — one whose processes are all stopped or gone is a likely leftover. It cannot tell you who made a workspace; the user arranges them by hand too.
run_processwrite actionstart_projectStart a user-registered project with its registered command, in its registered path. Neither can be changed. PREFER this over run_process: call list_projects first, and if a project is registered inside the checkout you are working in, start it here instead of running the command yourself. Do NOT use this from a git worktree for a project registered in the main checkout: it would start the main ch
stop_processStop an AI process by its exact display name. Prefer stop_project: it takes the id from list_projects, works for projects and AI processes alike, and the id is what every other tool uses. Use this only when you know an AI process by name alone. Does not apply to permanent projects. If no AI process has exactly that name, nothing is stopped, but the response still reads as sent.
stop_projectStop a running project or AI process by id (from list_projects). This is the stop tool to use for both kinds. The entry stays registered: a project can be started again with start_project. To remove an AI process instead of only stopping it, use delete_process. Returns once the stop is requested; call get_status to confirm.
update_projectwrite actionUpdate a kept (permanent) DevGlow project's metadata. Pass only the fields you want to change. Does NOT apply to temporary AI processes — for those, set fields at run_process time or have the user keep the process first.
workspace_runwrite action
Public scan report
scanner v0.1.10 · 2026-11-04 · same rubric, same numbers if you re-run it
- Code scan7 source files scanned25/25
- –Live reliabilityno gateway calls yet and no remote to proben/a
- –Tool poisoningtools not inspected (local package is not executed); not countedn/a
- Auth qualitylocal package, no credentials required12/15
- Maintenancelast push 12 days ago15/15
- Maintainer identitynamespace and repository owner differ; GitHub account older than a year; website matches verified namespace7/10
How devglow compares
Against every graded AI and agents MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 91/100
- higher than 88% of 4,460 AI and agents servers
- Tools
- 17
- the typical AI and agents server has 6
- Tools that change things
- 47%
- typical: 0%. Fewer write tools means less an agent can break
- Downloads a week
- 152
- more than 71% of 1,538 AI and agents servers with a package
What its tools let an agent do: 6 read · 1 create · 3 change · 3 delete · 4 run or deploy
Release history: 2 versions in the registry since 2026-08-11, latest 2026-10-23, typically one every 73 days.
Install devglow in Claude Code, Cursor or VS Code
Runs npx -y devglow-mcp on your machine. Read the scan report first; the gateway never runs local packages.
claude mcp add devglow -- npx -y devglow-mcp
What the publisher says
From the devglow repository's README, as published. We do not edit it. Read it on GitHub
devglow releases
Download the latest release from the Releases page.
Nothing above is checked by us. What we check is on the safety report.
devglow: common questions
- Is devglow MCP server safe?
- Yes, by our scan: it is graded A (91/100). Read the devglow safety report
- How do I install devglow?
- It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
- Does devglow need an API key?
- Not as far as the registry entry and our scan can tell: no credentials are declared or required.
- Is devglow maintained?
- The last commit was 12 days ago (2026-10-23). The latest release is v1.6.0.
- What can I use instead of devglow?
- Servers from other publishers that do the same job: agent-device MCP server, agent-device MCP server and Watch Skill MCP server.
- How does devglow compare with other AI and agents MCP servers?
- safety score 91/100 (higher than 88% of 4,460 AI and agents servers); tools 17 (the typical AI and agents server has 6); tools that change things 47% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.
Alternatives to devglow
Same job from other publishers: the closest match first, then the best rated.
- agent-deviceLet AI agents inspect, control, and debug real iOS, Android, desktop, and TV appsnot reviewedWidely usedA
- agent-deviceOpen source device layer for AI agents: control mobile, web, TV, and desktop apps over MCPnot reviewedWidely usedA
- Watch SkillWatch video and live sessions, keep timestamped evidence, and verify an agent's own work.not reviewedEstablishedB
- AKShareLet a model explore AKShare's 1000+ China market data functions on its own — search, inspect, callnot reviewedEstablishedA
- EmuloMine your AI coding session logs into a you.md profile with dated receipts, for an agent to read.not reviewedEstablishedA