Pipelines

A project’s pipelines: list them, read one or the latest, run one, wait for one to finish, retry, cancel and delete one, update its metadata, and read its variables and its JUnit test report or summary. Deleting a pipeline removes its jobs with it.

Pipeline schedules (with their variables, running one now and taking one over), pipeline trigger tokens (and running a pipeline with one), and resource groups, the locks that keep jobs from running at once, are actions of this group too.

Sample questions

• “Show the pipelines of project 42”
• “What is the status of the latest pipeline on main?”
• “Retry the failed pipeline”
• “Which tests failed in pipeline 1234?”

How to call it

• Dynamic, the default surface: call gitlab_execute_action with action set to the action’s ID, such as pipeline.cancel, and its parameters in params. gitlab_find_action finds an ID from a description of the task.
• Meta (GITLAB_MCP_TOOL_SURFACE=meta): call gitlab_pipeline with action set to the action’s name, such as cancel, and its parameters in params.
• Individual (GITLAB_MCP_TOOL_SURFACE=individual): call the action’s own tool, such as gitlab_pipeline_cancel, with its parameters as the arguments.

Availability

Every tier serves the whole group, on self-managed instances and on GitLab.com alike.

Read-only actions: 15 of 33, the ones a deployment in read-only mode keeps.

Actions

The description of each action, and of each of its parameters, is the text the server serves for it on the default surface, quoted as served. A destructive action runs only once confirmed, unless GITLAB_MCP_YOLO_MODE (or AUTOPILOT) skips that step: the dynamic surface needs confirm: true on gitlab_execute_action, and the other two take a confirm parameter or the client’s prompt (Destructive actions).

Action	Individual
pipeline.cancel	gitlab_pipeline_cancel
pipeline.create	gitlab_pipeline_create
pipeline.delete	gitlab_pipeline_delete
pipeline.get	gitlab_pipeline_get
pipeline.latest	gitlab_pipeline_latest
pipeline.list	gitlab_pipeline_list
pipeline.resource_group_edit	gitlab_edit_resource_group
pipeline.resource_group_get	gitlab_get_resource_group
pipeline.resource_group_list	gitlab_list_resource_groups
pipeline.resource_group_upcoming_jobs	gitlab_list_resource_group_upcoming_jobs
pipeline.retry	gitlab_pipeline_retry
pipeline.schedule_create	gitlab_pipeline_schedule_create
pipeline.schedule_create_variable	gitlab_pipeline_schedule_create_variable
pipeline.schedule_delete	gitlab_pipeline_schedule_delete
pipeline.schedule_delete_variable	gitlab_pipeline_schedule_delete_variable
pipeline.schedule_edit_variable	gitlab_pipeline_schedule_edit_variable
pipeline.schedule_get	gitlab_pipeline_schedule_get
pipeline.schedule_list	gitlab_pipeline_schedule_list
pipeline.schedule_list_triggered_pipelines	gitlab_pipeline_schedule_list_triggered_pipelines
pipeline.schedule_run	gitlab_pipeline_schedule_run
pipeline.schedule_take_ownership	gitlab_pipeline_schedule_take_ownership
pipeline.schedule_update	gitlab_pipeline_schedule_update
pipeline.test_report	gitlab_pipeline_test_report
pipeline.test_report_summary	gitlab_pipeline_test_report_summary
pipeline.trigger_create	gitlab_pipeline_trigger_create
pipeline.trigger_delete	gitlab_pipeline_trigger_delete
pipeline.trigger_get	gitlab_pipeline_trigger_get
pipeline.trigger_list	gitlab_pipeline_trigger_list
pipeline.trigger_run	gitlab_pipeline_trigger_run
pipeline.trigger_update	gitlab_pipeline_trigger_update
pipeline.update_metadata	gitlab_pipeline_update_metadata
pipeline.variables	gitlab_pipeline_variables
pipeline.wait	gitlab_pipeline_wait

pipeline.cancel

Cancel a running pipeline. Returns: updated pipeline metadata reflecting the canceled status. See also: pipeline.get, pipeline.retry, job.list_project.

• Meta-tool: gitlab_pipeline, action cancel
• Individual tool: gitlab_pipeline_cancel
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to act on
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.create

Create a new pipeline. Returns: created pipeline metadata including ID, status, and target ref. See also: pipeline.get, pipeline.wait, job.list_project.

• Meta-tool: gitlab_pipeline, action create
• Individual tool: gitlab_pipeline_create
• Tier: Free
• Behavior: writes, not idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
ref	string	yes	Branch or tag name to run the pipeline for
inputs	object	no	Pipeline input parameters keyed by input name. Values may be string, number, boolean, or array of strings, matching the inputs declared in .gitlab-ci.yml spec:inputs.
variables	object[]	no	Pipeline variables to set

pipeline.delete

Delete a pipeline and all its jobs. Returns: a success status and confirmation message. See also: pipeline.get, pipeline.list, pipeline.cancel.

• Meta-tool: gitlab_pipeline, action delete
• Individual tool: gitlab_pipeline_delete
• Tier: Free
• Behavior: writes, destructive (needs confirmation), idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to delete
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.get

Get one pipeline by ID. Returns: full pipeline metadata, status lifecycle fields, and links. See also: pipeline.list, pipeline.variables, job.list_project, pipeline.retry.

• Meta-tool: gitlab_pipeline, action get
• Individual tool: gitlab_pipeline_get
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to retrieve
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.latest

Get the latest pipeline for a ref. Returns: full pipeline metadata for the most recent run. See also: pipeline.get, pipeline.list, job.list_project.

• Meta-tool: gitlab_pipeline, action latest
• Individual tool: gitlab_pipeline_latest
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
created_after	string	no	Return pipelines created after date (ISO 8601 format). Applies to the list fallback.
created_before	string	no	Return pipelines created before date (ISO 8601 format). Applies to the list fallback.
name	string	no	Filter by pipeline name. Applies to the list fallback.
order_by	string (id, status, ref, updated_at, user_id)	no	Order by field (id, status, ref, updated_at, user_id). Applies to the list fallback.
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
page_token	string	no	Keyset pagination cursor: record id at which to fetch the next page, taken from the previous keyset response. Only used when pagination=‘keyset’.
pagination	string	no	Pagination method: ‘keyset’ for keyset-based pagination on large ordered result sets, or ‘offset’ (the default). Keyset avoids deep-offset cost.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.
ref	string	no	Branch or tag name to filter by (defaults to the default branch)
scope	string (running, pending, finished, branches, tags)	no	Filter by scope (running, pending, finished, branches, tags). Applies to the list fallback.
sha	string	no	Filter by commit SHA. Applies to the list fallback.
sort	string (asc, desc)	no	Sort direction (asc, desc). Applies to the list fallback.
source	string (api, chat, external, external_pull_request_event, merge_request_event, ondemand_dast_scan, ondemand_dast_validation, parent_pipeline, pipeline, push, schedule, security_orchestration_policy, trigger, web, webide)	no	Filter by pipeline source (api, chat, external, external_pull_request_event, merge_request_event, ondemand_dast_scan, ondemand_dast_validation, parent_pipeline, pipeline, push, schedule, security_orchestration_policy, trigger, web, webide). Applies to the list fallback.
status	string (created, waiting_for_resource, preparing, pending, running, success, failed, canceled, skipped, manual, scheduled)	no	Filter by status (created, waiting_for_resource, preparing, pending, running, success, failed, canceled, skipped, manual, scheduled). Applies to the list fallback.
updated_after	string	no	Return pipelines updated after date (ISO 8601 format). Applies to the list fallback.
updated_before	string	no	Return pipelines updated before date (ISO 8601 format). Applies to the list fallback.
username	string	no	Filter by username that triggered the pipeline. Applies to the list fallback.
yaml_errors	boolean	no	Return only pipelines with YAML errors. Applies to the list fallback.

pipeline.list

List project pipelines with filters and pagination. Returns: pipeline IDs, refs, statuses, source, and timing metadata. See also: pipeline.get, pipeline.latest, job.list_project.

• Meta-tool: gitlab_pipeline, action list
• Individual tool: gitlab_pipeline_list
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
created_after	string	no	Return pipelines created after date (ISO 8601 format, e.g. 2025-01-01T00:00:00Z)
created_before	string	no	Return pipelines created before date (ISO 8601 format, e.g. 2025-12-31T23:59:59Z)
name	string	no	Filter by pipeline name
order_by	string (id, status, ref, updated_at, user_id)	no	Order by field (id, status, ref, updated_at, user_id)
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
page_token	string	no	Keyset pagination cursor: record id at which to fetch the next page, taken from the previous keyset response. Only used when pagination=‘keyset’.
pagination	string	no	Pagination method: ‘keyset’ for keyset-based pagination on large ordered result sets, or ‘offset’ (the default). Keyset avoids deep-offset cost.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.
ref	string	no	Filter by branch or tag name
scope	string (running, pending, finished, branches, tags)	no	Filter by scope (running, pending, finished, branches, tags)
sha	string	no	Filter by commit SHA
sort	string (asc, desc)	no	Sort direction (asc, desc)
source	string (api, chat, external, external_pull_request_event, merge_request_event, ondemand_dast_scan, ondemand_dast_validation, parent_pipeline, pipeline, push, schedule, security_orchestration_policy, trigger, web, webide)	no	Filter by pipeline source (api, chat, external, external_pull_request_event, merge_request_event, ondemand_dast_scan, ondemand_dast_validation, parent_pipeline, pipeline, push, schedule, security_orchestration_policy, trigger, web, webide)
status	string (created, waiting_for_resource, preparing, pending, running, success, failed, canceled, skipped, manual, scheduled)	no	Filter by status (created, waiting_for_resource, preparing, pending, running, success, failed, canceled, skipped, manual, scheduled)
updated_after	string	no	Return pipelines updated after date (ISO 8601 format, e.g. 2025-01-01T00:00:00Z)
updated_before	string	no	Return pipelines updated before date (ISO 8601 format, e.g. 2025-12-31T23:59:59Z)
username	string	no	Filter by username that triggered the pipeline
yaml_errors	boolean	no	Return only pipelines with YAML errors

pipeline.resource_group_edit

Update the process mode of one CI resource group by key. Returns: the updated resource group ID, key, new process mode that controls how queued jobs sharing the resource group are serialized, and created and updated times. See also: pipeline.resource_group_get, pipeline.resource_group_list, pipeline.resource_group_upcoming_jobs.

• Meta-tool: gitlab_pipeline, action resource_group_edit
• Individual tool: gitlab_edit_resource_group
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Resource group key
process_mode	string (unordered, oldest_first, newest_first, newest_ready_first)	yes	Process mode: unordered, oldest_first, newest_first, or newest_ready_first
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.resource_group_get

Get one CI resource group in a project by key. Returns: the resource group ID, key, process mode (the concurrency mode that controls how jobs sharing the resource group are serialized), and created and updated times. See also: pipeline.resource_group_list, pipeline.resource_group_edit, pipeline.resource_group_upcoming_jobs.

• Meta-tool: gitlab_pipeline, action resource_group_get
• Individual tool: gitlab_get_resource_group
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Resource group key
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.resource_group_list

List the CI resource groups configured for a project. Returns: each resource group’s ID, key, process mode that controls how jobs sharing the group are serialized to limit pipeline concurrency, and created and updated times, with pagination metadata. See also: pipeline.resource_group_get, pipeline.resource_group_edit, pipeline.resource_group_upcoming_jobs.

• Meta-tool: gitlab_pipeline, action resource_group_list
• Individual tool: gitlab_list_resource_groups
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.

pipeline.resource_group_upcoming_jobs

List the upcoming CI jobs queued for one resource group by key. Returns: each pending job’s ID, name, status, stage, ref, tag and allow-failure flags, pipeline, web URL, and creation time, ordered as they will run under the resource group’s process mode, with pagination metadata. See also: pipeline.resource_group_get, pipeline.resource_group_list, job.list.

• Meta-tool: gitlab_pipeline, action resource_group_upcoming_jobs
• Individual tool: gitlab_list_resource_group_upcoming_jobs
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Resource group key
project_id	string/integer	yes	Project ID or URL-encoded path
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.

pipeline.retry

Retry failed and canceled jobs in a pipeline. Returns: updated pipeline metadata after re-queuing jobs. See also: pipeline.get, pipeline.cancel, job.list_project.

• Meta-tool: gitlab_pipeline, action retry
• Individual tool: gitlab_pipeline_retry
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to act on
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.schedule_create

Create a pipeline schedule. Returns: the created schedule with id, description, ref, cron, owner, and inputs. See also: pipeline.schedule_get, pipeline.schedule_update, pipeline.schedule_list.

• Meta-tool: gitlab_pipeline, action schedule_create
• Individual tool: gitlab_pipeline_schedule_create
• Tier: Free
• Behavior: writes, not idempotent

Parameter	Type	Mandatory	Description
cron	string	yes	Cron expression (e.g. 0 1 * * *)
description	string	yes	Schedule description
project_id	string/integer	yes	Project ID or URL-encoded path
ref	string	yes	Branch or tag to run the pipeline on
active	boolean	no	Whether the schedule is active (default: true)
cron_timezone	string	no	Cron timezone (e.g. UTC or America/New_York)
inputs	object[]	no	Pipeline inputs to seed on the schedule. Each entry has name and value

pipeline.schedule_create_variable

Create a pipeline schedule variable. Returns: the created variable with key, value, variable_type, and raw. See also: pipeline.schedule_edit_variable, pipeline.schedule_delete_variable.

• Meta-tool: gitlab_pipeline, action schedule_create_variable
• Individual tool: gitlab_pipeline_schedule_create_variable
• Tier: Free
• Behavior: writes, not idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Variable key
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID
value	string	yes	Variable value. Required when creating a schedule variable
variable_type	string (env_var, file)	no	Variable type: env_var (default) or file

pipeline.schedule_delete

Delete a pipeline schedule permanently. Returns: a success confirmation. See also: pipeline.schedule_get, pipeline.schedule_list.

• Meta-tool: gitlab_pipeline, action schedule_delete
• Individual tool: gitlab_pipeline_schedule_delete
• Tier: Free
• Behavior: writes, destructive (needs confirmation), idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID

pipeline.schedule_delete_variable

Delete a pipeline schedule variable. Returns: a success confirmation naming the key. See also: pipeline.schedule_create_variable, pipeline.schedule_edit_variable.

• Meta-tool: gitlab_pipeline, action schedule_delete_variable
• Individual tool: gitlab_pipeline_schedule_delete_variable
• Tier: Free
• Behavior: writes, destructive (needs confirmation), idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Variable key to delete
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID

pipeline.schedule_edit_variable

Edit a pipeline schedule variable. Returns: the updated variable with key, value, variable_type, and raw. See also: pipeline.schedule_create_variable, pipeline.schedule_delete_variable.

• Meta-tool: gitlab_pipeline, action schedule_edit_variable
• Individual tool: gitlab_pipeline_schedule_edit_variable
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
key	string	yes	Variable key to edit
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID
value	string	yes	New variable value. Required when editing a schedule variable
variable_type	string (env_var, file)	no	Variable type: env_var or file

pipeline.schedule_get

Get a single pipeline schedule by ID. Returns: the schedule with cron, ref, owner, last_pipeline, variables, and inputs. See also: pipeline.schedule_list, pipeline.schedule_update, pipeline.schedule_run.

• Meta-tool: gitlab_pipeline, action schedule_get
• Individual tool: gitlab_pipeline_schedule_get
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID

pipeline.schedule_list

List pipeline schedules in a project. Returns: schedules with description, ref, cron, owner, and pagination metadata. See also: pipeline.schedule_get, pipeline.schedule_create.

• Meta-tool: gitlab_pipeline, action schedule_list
• Individual tool: gitlab_pipeline_schedule_list
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
order_by	string	no	Column to order keyset-paginated results by (e.g. id)
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
page_token	string	no	Keyset pagination cursor: record id at which to fetch the next page, taken from the previous keyset response. Only used when pagination=‘keyset’.
pagination	string	no	Pagination method: ‘keyset’ for keyset-based pagination on large ordered result sets, or ‘offset’ (the default). Keyset avoids deep-offset cost.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.
scope	string (active, inactive)	no	Filter by scope: active or inactive
sort	string (asc, desc)	no	Sort order for keyset pagination: asc or desc

pipeline.schedule_list_triggered_pipelines

List pipelines triggered by a schedule. Returns: triggered pipelines with status, ref, sha, web URL, and pagination metadata. See also: pipeline.schedule_get, pipeline.schedule_run.

• Meta-tool: gitlab_pipeline, action schedule_list_triggered_pipelines
• Individual tool: gitlab_pipeline_schedule_list_triggered_pipelines
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID
order_by	string	no	Column to order results by (e.g. id)
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
page_token	string	no	Keyset pagination cursor: record id at which to fetch the next page, taken from the previous keyset response. Only used when pagination=‘keyset’.
pagination	string	no	Pagination method: ‘keyset’ for keyset-based pagination on large ordered result sets, or ‘offset’ (the default). Keyset avoids deep-offset cost.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.
sort	string (asc, desc)	no	Sort direction for triggered pipelines: asc or desc (default asc)

pipeline.schedule_run

Run a pipeline schedule immediately. Returns: the schedule’s current state after triggering. See also: pipeline.schedule_get, pipeline.schedule_list_triggered_pipelines.

• Meta-tool: gitlab_pipeline, action schedule_run
• Individual tool: gitlab_pipeline_schedule_run
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID

pipeline.schedule_take_ownership

Take ownership of a pipeline schedule. Returns: the schedule with the caller as owner. See also: pipeline.schedule_update, pipeline.schedule_run.

• Meta-tool: gitlab_pipeline, action schedule_take_ownership
• Individual tool: gitlab_pipeline_schedule_take_ownership
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID

pipeline.schedule_update

Update a pipeline schedule. Returns: the updated schedule with cron, ref, owner, and inputs. See also: pipeline.schedule_get, pipeline.schedule_take_ownership.

• Meta-tool: gitlab_pipeline, action schedule_update
• Individual tool: gitlab_pipeline_schedule_update
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or URL-encoded path
schedule_id	integer	yes	Pipeline schedule ID
active	boolean	no	Enable or disable the schedule
cron	string	no	Updated cron expression
cron_timezone	string	no	Updated cron timezone
description	string	no	Updated description
inputs	object[]	no	Pipeline inputs to set on the schedule. Each entry has name, value, and optional destroy
ref	string	no	Updated branch or tag

pipeline.test_report

Get a pipeline’s full test report. Returns: total/passed/failed/skipped counts and per-suite breakdown. See also: pipeline.test_report_summary, pipeline.get, job.list_project.

• Meta-tool: gitlab_pipeline, action test_report
• Individual tool: gitlab_pipeline_test_report
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to retrieve
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.test_report_summary

Get a pipeline’s test report summary. Returns: aggregate test totals and per-suite summaries with build IDs. See also: pipeline.test_report, pipeline.get, job.list_project.

• Meta-tool: gitlab_pipeline, action test_report_summary
• Individual tool: gitlab_pipeline_test_report_summary
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to retrieve
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.trigger_create

Create a pipeline trigger token. Returns: the created trigger with id, description, token secret, and owner. See also: pipeline.trigger_run, pipeline.trigger_list, pipeline.trigger_delete.

• Meta-tool: gitlab_pipeline, action trigger_create
• Individual tool: gitlab_pipeline_trigger_create
• Tier: Free
• Behavior: writes, not idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
description	string	no	Trigger token description

pipeline.trigger_delete

Delete a pipeline trigger token. Returns: a success confirmation. See also: pipeline.trigger_get, pipeline.trigger_create.

• Meta-tool: gitlab_pipeline, action trigger_delete
• Individual tool: gitlab_pipeline_trigger_delete
• Tier: Free
• Behavior: writes, destructive (needs confirmation), idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
trigger_id	integer	yes	Pipeline trigger ID

pipeline.trigger_get

Get a single pipeline trigger token. Returns: the trigger with id, description, token, owner, and timestamps. See also: pipeline.trigger_list, pipeline.trigger_update, pipeline.trigger_delete.

• Meta-tool: gitlab_pipeline, action trigger_get
• Individual tool: gitlab_pipeline_trigger_get
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
trigger_id	integer	yes	Pipeline trigger ID

pipeline.trigger_list

List a project’s pipeline trigger tokens. Returns: trigger tokens with id, description, owner, last-used time, and pagination metadata. See also: pipeline.trigger_get, pipeline.trigger_create, pipeline.trigger_run.

• Meta-tool: gitlab_pipeline, action trigger_list
• Individual tool: gitlab_pipeline_trigger_list
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
order_by	string	no	Field by which to order results when using keyset pagination (e.g. id).
page	integer	no	Page number to fetch, 1-based. Defaults to 1. Use the next_page field from the previous response to paginate forward.
page_token	string	no	Keyset pagination cursor: record id at which to fetch the next page, taken from the previous keyset response. Only used when pagination=‘keyset’.
pagination	string	no	Pagination method: ‘keyset’ for keyset-based pagination on large ordered result sets, or ‘offset’ (the default). Keyset avoids deep-offset cost.
per_page	integer	no	Items per page. Defaults to 20, minimum 1, maximum 100. Use 100 to minimize round trips when the result set is large.
sort	string (asc, desc)	no	Sort order: asc or desc. Defaults to the API ordering.

pipeline.trigger_run

Trigger a pipeline using a trigger token. Returns: the created pipeline with id, status, ref, sha, source, user, detailed_status, and web URL. See also: pipeline.trigger_list, pipeline.get.

• Meta-tool: gitlab_pipeline, action trigger_run
• Individual tool: gitlab_pipeline_trigger_run
• Tier: Free
• Behavior: writes, not idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
inputs	object	no	Map of pipeline input name to value (string, number, boolean, or array of strings) for inputs declared in the pipeline spec.
ref	string	no	Branch or tag name to run pipeline on
token	string	no	Pipeline trigger token
variables	object	no	Map of CI/CD variable name to value injected into the triggered pipeline.

pipeline.trigger_update

Update a pipeline trigger token’s description. Returns: the updated trigger with id, description, token, and owner. See also: pipeline.trigger_get, pipeline.trigger_delete.

• Meta-tool: gitlab_pipeline, action trigger_update
• Individual tool: gitlab_pipeline_trigger_update
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
project_id	string/integer	yes	Project ID or path
trigger_id	integer	yes	Pipeline trigger ID
description	string	no	New trigger token description

pipeline.update_metadata

Update a pipeline’s name. Returns: updated pipeline metadata reflecting the new name. See also: pipeline.get, pipeline.list, pipeline.create.

• Meta-tool: gitlab_pipeline, action update_metadata
• Individual tool: gitlab_pipeline_update_metadata
• Tier: Free
• Behavior: writes, idempotent

Parameter	Type	Mandatory	Description
name	string	yes	New pipeline name
pipeline_id	integer	yes	Pipeline ID to update
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.variables

List a pipeline’s runtime variables. Returns: variable keys, values, types, and whether each value is raw (not expanded). See also: pipeline.get, pipeline.create, job.list_project.

• Meta-tool: gitlab_pipeline, action variables
• Individual tool: gitlab_pipeline_variables
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID to retrieve
project_id	string/integer	yes	Project ID or URL-encoded path

pipeline.wait

Use only to poll an existing pipeline_id until a terminal status. For merge when pipeline succeeds, use merge_request.merge with auto_merge=true instead.

• Meta-tool: gitlab_pipeline, action wait
• Individual tool: gitlab_pipeline_wait
• Tier: Free
• Behavior: read-only, idempotent

Parameter	Type	Mandatory	Description
pipeline_id	integer	yes	Pipeline ID returned by pipeline.list, pipeline.get, pipeline.latest, or merge_request.pipelines. Do not use merge_request_iid.
project_id	string/integer	yes	Project ID or URL-encoded path
fail_on_error	boolean	no	Return isError when pipeline ends in failed/canceled status (default true)
interval_seconds	integer	no	Polling interval in seconds (5-60, default 10)
timeout_seconds	integer	no	Maximum wait time in seconds (1-3600, default 300)