MCP Broken?

[sjpritchard]

I'm having issues with MCP no longer working in my RooCode instance using Gemini 2.5. I did a reset of RooCode seetings and a fresh install of RooFlow. During install of RooFlowI get this:

Processing target prompt files...
Processing: /home/xxx/.roo/system-prompt-flow-code
  Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-code.
  Completed: system-prompt-flow-code
Processing: /home/xxx/.roo/system-prompt-flow-ask
  Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-ask.
  Completed: system-prompt-flow-ask
Processing: /home/xxx/.roo/system-prompt-flow-debug
  Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-debug.
  Completed: system-prompt-flow-debug
Processing: /home/xxx/.roo/system-prompt-flow-architect
  Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-architect.
  Completed: system-prompt-flow-architect
Processing complete.
--- RooFlow config setup complete ---

And when I ask RooCode if it can see any tools:

Goal: Re-check for MCP tools based on the user's follow-up question.
Analysis: The user asked again if I can see MCP tools. My ability to see/use MCP tools depends on the [CONNECTED_MCP_SERVERS] list provided in my configuration. I need to re-examine that part of my configuration.
MCP Information Source: Checking the mcp_servers_info section of my system prompt again.
Current State: The [CONNECTED_MCP_SERVERS] section remains empty. There are still no MCP servers listed as connected.
Task Status: The answer is the same as before. The task (checking for MCP tools) is complete based on my current configuration.
Tool Selection: Use attempt_completion as the task is complete.
Result: State clearly that, based on my current configuration, no MCP servers are connected.
Command: No command is needed.

The native modes work fine. Issue is the same with Sonnet 3.7 too.

[GreatScottyMac]

In the README.md look for the section: "Importing Connected MCP Server Tools (Optional)"

…

On Thu, May 1, 2025, 11:09 PM Steve Pritchard ***@***.***> wrote: *sjpritchard* created an issue (GreatScottyMac/RooFlow#58) <#58> I'm having issues with MCP no longer working in my RooCode instance using Gemini 2.5. I did a reset of RooCode seetings and a fresh install of RooFlow. During install of RooFlowI get this: Processing target prompt files... Processing: /home/xxx/.roo/system-prompt-flow-code Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-code. Completed: system-prompt-flow-code Processing: /home/xxx/.roo/system-prompt-flow-ask Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-ask. Completed: system-prompt-flow-ask Processing: /home/xxx/.roo/system-prompt-flow-debug Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-debug. Completed: system-prompt-flow-debug Processing: /home/xxx/.roo/system-prompt-flow-architect Placeholder/Block found but no MCP content generated. Skipping injection in system-prompt-flow-architect. Completed: system-prompt-flow-architect Processing complete. --- RooFlow config setup complete --- And when I ask RooCode if it can see any tools: Goal: Re-check for MCP tools based on the user's follow-up question. Analysis: The user asked again if I can see MCP tools. My ability to see/use MCP tools depends on the [CONNECTED_MCP_SERVERS] list provided in my configuration. I need to re-examine that part of my configuration. MCP Information Source: Checking the mcp_servers_info section of my system prompt again. Current State: The [CONNECTED_MCP_SERVERS] section remains empty. There are still no MCP servers listed as connected. Task Status: The answer is the same as before. The task (checking for MCP tools) is complete based on my current configuration. Tool Selection: Use attempt_completion as the task is complete. Result: State clearly that, based on my current configuration, no MCP servers are connected. Command: No command is needed. The native modes work fine. Issue is the same with Sonnet 3.7 too. — Reply to this email directly, view it on GitHub <#58>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BNSRHLOI2AN23Z4XH646HWL24LOVDAVCNFSM6AAAAAB4JDNIUOVHI2DSMVQWIX3LMV43ASLTON2WKOZTGAZTKMBYHAZTENY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[sjpritchard]

Ah, apologies, I missed that. Thanks for your help!

[sjpritchard]

I RTFM'ed this time, however the MCP server is missing tools and resources in the generated prompts.

Original system prompt excerpt:

`## context7 (npx -y @upstash/context7-mcp@latest)

Available Tools

• resolve-library-id: Required first step: Resolves a general package name into a Context7-compatible library ID. Must be called before using 'get-library-docs' to retrieve a valid Context7-compatible library ID.
  Input Schema:
  {
  "type": "object",
  "properties": {
  "libraryName": {
  "type": "string",
  "description": "Library name to search for and retrieve a Context7-compatible library ID."
  }
  },
  "required": [
  "libraryName"
  ],
  "additionalProperties": false,
  "$schema": "http://json-schema.org/draft-07/schema#"
  }

• get-library-docs: Fetches up-to-date documentation for a library. You must call 'resolve-library-id' first to obtain the exact Context7-compatible library ID required to use this tool.
  Input Schema:
  {
  "type": "object",
  "properties": {
  "context7CompatibleLibraryID": {
  "type": "string",
  "description": "Exact Context7-compatible library ID (e.g., 'mongodb/docs', 'vercel/nextjs') retrieved from 'resolve-library-id'."
  },
  "topic": {
  "type": "string",
  "description": "Topic to focus documentation on (e.g., 'hooks', 'routing')."
  },
  "tokens": {
  "type": "number",
  "description": "Maximum number of tokens of documentation to retrieve (default: 5000). Higher values provide more context but consume more tokens."
  }
  },
  "required": [
  "context7CompatibleLibraryID"
  ],
  "additionalProperties": false,
  "$schema": "http://json-schema.org/draft-07/schema#"
  }`

Install script:

MCP YAML generated successfully. Processing target prompt files... Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-code Injecting/Overwriting MCP block in system-prompt-flow-code... Completed: system-prompt-flow-code Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-ask Injecting/Overwriting MCP block in system-prompt-flow-ask... Completed: system-prompt-flow-ask Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-debug Injecting/Overwriting MCP block in system-prompt-flow-debug... Completed: system-prompt-flow-debug Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-architect Injecting/Overwriting MCP block in system-prompt-flow-architect... Completed: system-prompt-flow-architect Processing complete.

system-prompt-flow-ask:

# MCP Server list injected by script servers: - name: context7 command: npx -y @upstash/context7-mcp@latest description: '' tools: [] resources: []

[GreatScottyMac]

I've only tested it on a few MCP servers so far. For now, you could just copy and paste the MCP section directly from the Roo Code prompt to the RooFlow prompts, or use the native modes.

I should be able to troubleshoot it in a day or two.

I'm curious to see if there's any difference in performance with the MCP tools presented to the LLMs in YAML.

[redjumper25]

I'm having an error with this on arm64 as the generate_mcp_yaml binary is compiled only for x86-64.

./install_rooflow.sh: line 76: ./generate_mcp_yaml: cannot execute binary file

Below is some information about the binary

username@computer workspace % file generate_mcp_yaml
generate_mcp_yaml: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 3.2.0, BuildID[sha1]=4457c553620a25b2fdf6b49658b64edb4cd3f053, stripped

Also, is there any reason why this needs to be a compiled binary?

[GreatScottyMac]

I RTFM'ed this time, however the MCP server is missing tools and resources in the generated prompts.

Original system prompt excerpt:

`## context7 (npx -y @upstash/context7-mcp@latest)

Available Tools

• resolve-library-id: Required first step: Resolves a general package name into a Context7-compatible library ID. Must be called before using 'get-library-docs' to retrieve a valid Context7-compatible library ID.
  Input Schema:
  {
  "type": "object",
  "properties": {
  "libraryName": {
  "type": "string",
  "description": "Library name to search for and retrieve a Context7-compatible library ID."
  }
  },
  "required": [
  "libraryName"
  ],
  "additionalProperties": false,
  "$schema": "http://json-schema.org/draft-07/schema#"
  }
• get-library-docs: Fetches up-to-date documentation for a library. You must call 'resolve-library-id' first to obtain the exact Context7-compatible library ID required to use this tool.
  Input Schema:
  {
  "type": "object",
  "properties": {
  "context7CompatibleLibraryID": {
  "type": "string",
  "description": "Exact Context7-compatible library ID (e.g., 'mongodb/docs', 'vercel/nextjs') retrieved from 'resolve-library-id'."
  },
  "topic": {
  "type": "string",
  "description": "Topic to focus documentation on (e.g., 'hooks', 'routing')."
  },
  "tokens": {
  "type": "number",
  "description": "Maximum number of tokens of documentation to retrieve (default: 5000). Higher values provide more context but consume more tokens."
  }
  },
  "required": [
  "context7CompatibleLibraryID"
  ],
  "additionalProperties": false,
  "$schema": "http://json-schema.org/draft-07/schema#"
  }`

Install script:

MCP YAML generated successfully. Processing target prompt files... Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-code Injecting/Overwriting MCP block in system-prompt-flow-code... Completed: system-prompt-flow-code Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-ask Injecting/Overwriting MCP block in system-prompt-flow-ask... Completed: system-prompt-flow-ask Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-debug Injecting/Overwriting MCP block in system-prompt-flow-debug... Completed: system-prompt-flow-debug Processing: /home/steve/simcortex/reforger_doxygen_converter/.roo/system-prompt-flow-architect Injecting/Overwriting MCP block in system-prompt-flow-architect... Completed: system-prompt-flow-architect Processing complete.

system-prompt-flow-ask:

# MCP Server list injected by script servers: - name: context7 command: npx -y @upstash/context7-mcp@latest description: '' tools: [] resources: []

This was my result, I was unable to reproduce the output that you received:

• name: upstash
  command: npx -y @upstash/mcp-server run @gmail.com
  description: ''
  tools:
  - name: redis_database_create_new
  description: 'Create a new Upstash redis database. NOTE: Ask user for the region and name of the database. NOTE: Don''t show the database ID from the response to the user unless explicitly asked or needed.'
  input_schema:
  type: object
  properties:
  name:
  type: string
  description: Name of the database.
  primary_region:
  type: string
  enum:
  - us-east-1
  - us-west-1
  - us-west-2
  - eu-west-1
  - eu-central-1
  - ap-southeast-1
  - ap-southeast-2
  - sa-east-1
  description: Primary Region of the Global Database.
  read_regions:
  type: array
  items:
  type: string
  enum:
  - us-east-1
  - us-west-1
  - us-west-2
  - eu-west-1
  - eu-central-1
  - ap-southeast-1
  - ap-southeast-2
  - sa-east-1
  description: Array of read regions of the db
  required:
  - name
  - primary_region
  - name: redis_database_delete
  description: Delete an Upstash redis database.
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to delete.
  required:
  - database_id
  - name: redis_database_list_databases
  description: 'List all Upstash redis databases. Only their names and ids. NOTE: Don''t show the database ID from the response to the user unless explicitly asked or needed.'
  input_schema:
  type: object
  properties: {}
  - name: redis_database_get_details
  description: 'Get further details of a specific Upstash redis database. Includes all details of the database including usage statistics. db_disk_threshold: Total disk usage limit. db_memory_threshold: Maximum memory usage. db_daily_bandwidth_limit: Maximum daily network bandwidth usage. db_request_limit: Total number of commands allowed. All sizes are in bytes NOTE: Don''t show the database ID from the response to the user unless explicitly asked or needed.'
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to get details for.
  required:
  - database_id
  - name: redis_database_update_regions
  description: Update the read regions of an Upstash redis database.
  input_schema:
  type: object
  properties:
  id:
  type: string
  description: The ID of your database.
  read_regions:
  type: array
  items:
  type: string
  enum:
  - us-east-1
  - us-west-1
  - us-west-2
  - eu-west-1
  - eu-central-1
  - ap-southeast-1
  - ap-southeast-2
  - sa-east-1
  description: 'Array of the new read regions of the database. This will replace the old regions array. Available regions: us-east-1, us-west-1, us-west-2, eu-west-1, eu-central-1, ap-southeast-1, ap-southeast-2, sa-east-1'
  required:
  - id
  - read_regions
  - name: redis_database_reset_password
  description: Reset the password of an Upstash redis database.
  input_schema:
  type: object
  properties:
  id:
  type: string
  description: The ID of your database.
  required:
  - id
  - name: redis_database_get_usage_last_5_days
  description: 'Get PRECISE command count and bandwidth usage statistics of an Upstash redis database over the last 5 days. This is a precise stat, not an average. NOTE: Ask user first if they want to see stats for each database seperately or just for one.'
  input_schema:
  type: object
  properties:
  id:
  type: string
  description: The ID of your database.
  required:
  - id
  - name: redis_database_get_stats
  description: 'Get SAMPLED usage statistics of an Upstash redis database over a period of time (1h, 3h, 12h, 1d, 3d, 7d). Use this to check for peak usages and latency problems. Includes: read_latency_mean, write_latency_mean, keyspace, throughput (cmds/sec), diskusage NOTE: If the user does not specify which stat to get, use throughput as default.'
  input_schema:
  type: object
  properties:
  id:
  type: string
  description: The ID of your database.
  period:
  type: string
  enum:
  - 1h
  - 3h
  - 12h
  - 1d
  - 3d
  - 7d
  description: The period of the stats.
  type:
  anyOf:
  - type: string
  const: read_latency_mean
  - type: string
  const: write_latency_mean
  - type: string
  const: keyspace
  description: Number of keys in db
  - type: string
  const: throughput
  description: commands per second (sampled), calculate area for estimated count
  - type: string
  const: diskusage
  description: Current disk usage in bytes
  description: The type of stat to get
  required:
  - id
  - period
  - type
  - name: redis_database_create_backup
  description: 'Create a backup of a specific Upstash redis database. NOTE: Ask user to choose a name for the backup'
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to create a backup for.
  backup_name:
  type: string
  description: A name for the backup.
  required:
  - database_id
  - backup_name
  - name: redis_database_delete_backup
  description: Delete a backup of a specific Upstash redis database.
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to delete a backup from.
  backup_id:
  type: string
  description: The ID of the backup to delete.
  required:
  - database_id
  - backup_id
  - name: redis_database_restore_backup
  description: Restore a backup of a specific Upstash redis database. A backup can only be restored to the same database it was created from.
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to restore a backup to.
  backup_id:
  type: string
  description: The ID of the backup to restore.
  required:
  - database_id
  - backup_id
  - name: redis_database_list_backups
  description: List all backups of a specific Upstash redis database.
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to list backups for.
  required:
  - database_id
  - name: redis_database_set_daily_backup
  description: Enable or disable daily backups for a specific Upstash redis database.
  input_schema:
  type: object
  properties:
  database_id:
  type: string
  description: The ID of the database to enable or disable daily backups for.
  enable:
  type: boolean
  description: Whether to enable or disable daily backups.
  required:
  - database_id
  - enable
  - name: redis_database_run_single_redis_command
  description: 'Run a single Redis command on a specific Upstash redis database. NOTE: For discovery, use SCAN over KEYS. Use TYPE to get the type of a key. NOTE: SCAN cursor [MATCH pattern] [COUNT count] [TYPE type]'
  input_schema:
  type: object
  properties:
  database_rest_url:
  type: string
  description: 'The REST URL of the database. Example: https://.upstash.io'
  database_rest_token:
  type: string
  description: The REST token of the database.
  command:
  type: array
  items:
  type: string
  description: 'The Redis command to run. Example: [''SET'', ''foo'', ''bar'', ''EX'', 100]'
  required:
  - database_rest_url
  - database_rest_token
  - command
  - name: redis_database_run_multiple_redis_commands
  description: Run multiple Redis commands on a specific Upstash redis database
  input_schema:
  type: object
  properties:
  database_rest_url:
  type: string
  description: 'The REST URL of the database. Example: https://.upstash.io'
  database_rest_token:
  type: string
  description: The REST token of the database.
  commands:
  type: array
  items:
  type: array
  items:
  type: string
  description: 'The Redis commands to run. Example: [[''SET'', ''foo'', ''bar''], [''GET'', ''foo'']]'
  required:
  - database_rest_url
  - database_rest_token
  - commands
  - name: timestamps_to_date
  description: Use this tool to convert a timestamp to a human-readable date
  input_schema:
  type: object
  properties:
  timestamps:
  type: array
  items:
  type: number
  description: Array of timestamps to convert
  required:
  - timestamps
  resources: []

End MCP Server list

[GreatScottyMac]

I'm having an error with this on arm64 as the generate_mcp_yaml binary is compiled only for x86-64.

./install_rooflow.sh: line 76: ./generate_mcp_yaml: cannot execute binary file

Below is some information about the binary

username@computer workspace % file generate_mcp_yaml generate_mcp_yaml: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 3.2.0, BuildID[sha1]=4457c553620a25b2fdf6b49658b64edb4cd3f053, stripped

Also, is there any reason why this needs to be a compiled binary?

My intention was to eliminate python dependency for end-users but as you point out, this creates compatibility issues. I will revert the binary back into a python script with dependency checks.