check-mcp
Skill11 starsChanged 3 months agoMITView on GitHub
Diagnose why PAPI tools are missing or failing. Use when the user says "check mcp", "is papi connected", "papi isn't working", "mcp status", or when PAPI tools are expected but do not appear in the available tools list.
In this repository14 other files it gives its agents.
- llms.txtllms.txt
- deployment-completeness-auditSkill
- designSkill
- papi-advancedSkill
- papi-autoSkill
- papi-buildSkill
- papi-ideaSkill
- papi-integration-recipeSkill
- papi-planSkill
- papi-strategySkill
- papi-verifySkill
- pr-reviewerSkill
- security-auditSkill
- usage-reporterSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 2 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Look at the available tools list for PAPI tools such as orient, board_view and plan.
- Very specific: line 24
If PAPI tools are present, skip to Step 3.
- Distinctive: line 28
Check the client's MCP configuration.
- Very specific: line 30
Claude Code: run claude mcp list. To install, run /plugin marketplace add getpapi/papi then /plugin install papi@papi, or claude mcp add …
- Very specific: line 34
VS Code, Windsurf, Codex and others: check that client's MCP config for a papi entry pointing at https://mcp.getpapi.ai/mcp.
- Very specific: line 38
If papi is absent, PAPI is not installed.
- Very specific: line 41
If papi is listed but shows an error or "needs authentication", go to Step 3.
- Very specific: line 50
Claude Code: run /mcp, choose papi, pick Authenticate.
- Distinctive: line 55
If the user has no account, they create one at https://getpapi.ai.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Is PAPI connected?
PAPI runs as a hosted MCP server at https://mcp.getpapi.ai/mcp. Nothing runs on the
user's machine, so a failure is one of three things: the server is not registered, the
user is not signed in, or the account has no project bound.
Work through the steps in order and stop at the first one that fails. The checks are the same in every client; only the commands differ, so use the ones for the tool you are running in.
Step 1: Are the tools there?
Look at the available tools list for PAPI tools such as orient, board_view and plan.
Tool names vary by how PAPI was installed. Installed as a plugin in Claude Code they are
namespaced mcp__plugin_papi_papi__orient; added directly they are mcp__papi__orient;
other clients use their own prefixes. All of them mean the same server.
If PAPI tools are present, skip to Step 3. If none are, go to Step 2.
Step 2: Is the server registered?
Check the client's MCP configuration.
- Claude Code: run
claude mcp list. To install, run/plugin marketplace add getpapi/papithen/plugin install papi@papi, orclaude mcp add --transport http papi https://mcp.getpapi.ai/mcp. - Cursor: open Settings, then MCP, and look for
papi. Config lives in~/.cursor/mcp.jsonor.cursor/mcp.jsonin the project. - VS Code, Windsurf, Codex and others: check that client's MCP config for a
papientry pointing athttps://mcp.getpapi.ai/mcp. Per-client instructions are at https://getpapi.ai/docs/install.
If papi is absent, PAPI is not installed. Add it, then restart the client, since most
clients only read MCP config at startup.
If papi is listed but shows an error or "needs authentication", go to Step 3.
Step 3: Is the user signed in?
The hosted server requires a PAPI account. Registering the server does not sign anyone in, and this is the most common cause of "PAPI is installed but nothing works".
Sign-in is interactive and cannot be done on the user's behalf.
- Claude Code: run
/mcp, choose papi, pick Authenticate. A browser tab opens. - Other clients: the client prompts for OAuth on first use. If it never does, the client may not support OAuth discovery. Fall back to an API key from the dashboard's Connect panel, set as a bearer token in the MCP config. See https://getpapi.ai/docs/install.
If the user has no account, they create one at https://getpapi.ai. The MCP server authenticates against the website account, so there is no separate credential.
Step 4: Is a project bound?
Call orient.
- Returns cycle state: PAPI is healthy. Report and stop.
- Asks which project to use: the account has more than one project and this connection has
no default. Put the list to the user, then call again with their choice and run
project_switchto make it stick. - Reports an empty board or no cycles: the account is connected but this project has never
been set up. Call
setup.
Step 5: Report
Close with exactly one of these, and end on the single next action rather than a list of possibilities.
- Healthy: registered, signed in, project bound. Name the current cycle.
- Not signed in: registered but signed out. Next action is the sign-in step for this client.
- Not installed: no
papiserver registered. Next action is the install step for this client. - No project: connected, but nothing set up here. Next action is
setup.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

