- Category: model context protocol
- Auth: API_KEY
- Composio-managed OAuth available? N/A
- Tools: 23
- Triggers: 0
- Slug:
DEVIN_MCP - Version: 20261030_00
Tools
Ask wiki question
Ask wiki question
Devin automation manage
Devin automation manage
Slug:
DEVIN_MCP_DEVIN_AUTOMATION_MANAGEManage Devin automations.Automations run Devin in response to events (GitHub activity, Slack messages, Linear updates, schedules, incoming webhooks). Each automation has triggers (which events fire it, with optional conditions and replies) and actions (what to do, e.g. start a session with a prompt). Single-automation results include the webapp URL.Call action=“schemas” first when you build a new automation. It returns every supported trigger event type with its condition fields and replies, the RRULE and Slack rules, the action shapes, the config groups, and the validation constraints. Use action=“validate_create” or action=“validate_update” for a dry run that needs no approval.On create, follow “schemas” for session_settings.net_policy: when it reports a Git Manager starting policy, set it plus only the required hosts, even for tasks that need no external services; when a security profile governs the network, omit net_policy so sessions inherit the profile’s allowlist, and pick the tools.mcp_servers the task needs from the servers the profile allows. Use null only for explicitly requested unrestricted access. Omit session_settings for existing-session reminders without auto_create.On “update”, omitted top-level params stay unchanged. Whether a passed config group merges or replaces is per-organization: action=“schemas” reports it under “Update semantics”. When in doubt, fetch with “get” and include every field you want to keep.Input Parameters
Human-readable name for the automation (1-500 characters). Required for ‘create’ and ‘validate_create’.
Opaque cursor for ‘list’/‘preflight_events’ pagination (from a previous response’s end_cursor).
Tool grants for the sessions the automation spawns (slack_dm_scope goes inside this group, never as a top-level argument).
string ("list" | "get" | "schemas" | "templates" | "create" | "validate_create" | "update" | "validate_update" | "delete" | "run" | "preflight_get" | "preflight_samples" | "preflight_events" | "preflight_event" | "preflight_test" | "preflight_run")
required
‘templates’ lists the built-in templates whose template_id ‘create’ accepts. ‘run’ fires an enabled automation now and bypasses its trigger conditions. ‘validate_create’ and ‘validate_update’ dry-run the payload without approval. Preflight actions (only when ‘schemas’ reports it available): ‘preflight_get’ (saved script and settings), ‘preflight_samples’ (canned sample events to test against before any real event exists; with sample_event_type, that sample’s $EVENT_FILE), ‘preflight_events’ (an automation’s recent matched events), ‘preflight_event’ (one event’s $EVENT_FILE), ‘preflight_test’ (run a script, saved or not, on a real VM against a sample or a recent event; starts no actions; without automation_id it is a draft run for an automation that does not exist yet), ‘preflight_run’ (poll a test run’s outcome and logs). Save the script through the ‘preflight’ parameter of create/update.
For ‘run’ only: free-text stand-in for the triggering event, included in the event payload as additional context (max 10000 chars). Mutually exclusive with event_data.
Identity the triggered sessions run under. REQUIRED for ‘create’, no default. Choose it together with tools.mcp_servers. On ‘update’, omitted means unchanged.
What the automation does. Required for ‘create’. Replaces all existing actions on ‘update’. On a draft ‘preflight_test’ (no automation_id) pass the actions you will create with, so the draft applies the same default network policy as ‘create’. Most automations use one start_session action. Code scans use start_code_scan or scan_new_commits instead. Long prompts render as BEGIN/END PROMPT <label> blocks: extract only the text between those two lines into a local file, edit it, and update with the full actions array and ‘prompt’: ‘file:///absolute/path’ (Devin’s client resolves file URIs; other MCP clients must pass the text inline). Never retype a long prompt.
Enable or disable the automation on ‘create’/‘update’; on ‘list’, filter to automations in that enabled state.
Org-visible string key/value labels (max 16 pairs; keys <= 32 chars, values <= 128). Update behavior follows the org’s update semantics from action=“schemas”. On ‘list’, filters to automations whose metadata matches every given pair (null values are invalid there).
Events that dispatch the automation. Required for ‘create’. Replaces all existing triggers on ‘update’. Schedule triggers use an RFC 5545 RRULE condition with the ‘recurrence’ operator, e.g. {‘event_type’: ‘schedule:recurring’, ‘conditions’: {‘any’: [{‘all’: [{‘field’: ‘rrule’, ‘operator’: ‘recurrence’, ‘value’: ‘DTSTART;TZID=America/Los_Angeles:19700101T000000\nRRULE:FREQ=DAILY;BYHOUR=6;BYMINUTE=0’}]}]}}. Read the full RRULE rules and examples from action=‘schemas’.
For ‘create’/‘update’ and their validate_ dry-runs: the Preflight check, a script that runs before the actions and decides per trigger event whether to proceed, skip, or fan out. For ‘preflight_test’: the script to run (saved or not). Pass the source as ‘file:///absolute/path’. On update the whole group replaces the stored one; omitted keeps it. Preflight actions: ‘preflight_get’ (saved script), ‘preflight_events’ (recent matched trigger events), ‘preflight_event’ (one event’s full $EVENT_FILE document), ‘preflight_test’ (run a script against a recent event on a real VM; starts no actions), ‘preflight_run’ (poll a test run’s outcome and logs). Available only when action=‘schemas’ reports it.
For ‘run’ only: structured stand-in for the event data the real source would have sent. Mutually exclusive with prompt.
Filter ‘list’ results by agent type; multiple values match automations containing at least one action of any given type (OR). Values: ‘new_session’ (start_session), ‘long_running’ (message_session), ‘triage_session’ (auto-triage), ‘auto_triage’ (legacy monitor_session), ‘code_scan’ (start_code_scan / scan_new_commits), ‘remediate_finding’, ‘incident_session’.
The automation ID. Required for ‘get’, ‘update’, ‘validate_update’, ‘delete’, ‘run’, ‘preflight_get’, ‘preflight_events’, and ‘preflight_event’. Optional for ‘preflight_test’/‘preflight_run’: omit it to test a script for an automation that does not exist yet (a draft run against a sample event), then pass the same ‘preflight’ group to ‘create’.
For ‘preflight_event’ (required) and ‘preflight_test’ (optional; default = the most recent matched event that is not a manual run): a queued_event_id from ‘preflight_events’. Mutually exclusive with sample_event_type.
Security-profile binding: {‘profile_id’: ’…’} pins that profile, {‘profile_id’: null} records an explicit opt-out, omitted inherits the org/enterprise default. Writing it requires the security-profile management permission and is rejected from a Devin session: there, omit it and tell the user to pick the profile in the automation’s editor in the webapp and, when that profile has a network policy, to turn off the automation’s Network policy there so sessions use the profile’s allowlist. An opt-out cannot escape a mandatory profile. Reads return {‘selection’: ‘inherit’|‘none’|‘profile’, ‘profile_id’, ‘warnings’: […]}. Warnings are non-blocking configuration conflicts; show them to the user.
For ‘preflight_samples’ (optional: show that sample’s full $EVENT_FILE document) and ‘preflight_test’ (test against the canned sample for this trigger event type, e.g. ‘github:pull_request’; required when automation_id is omitted). Mutually exclusive with queued_event_id.
For ‘preflight_test’ with sample_event_type ‘webhook:incoming’ only: the JSON body the user’s system will POST to the webhook, used in place of the canned body (webhook payloads are user-defined; every other source’s shape is fixed and the canned sample is authoritative). Ask the user for a real example. At most 64KB.
Output
Devin billing tag manage
Devin billing tag manage
Slug:
DEVIN_MCP_DEVIN_BILLING_TAG_MANAGEManage Devin billing tags (groupings of Devin sessions for usage tracking) with a single action-based tool: list/get/create tags, assign sessions one at a time or in bulk (up to 500 per call, sent in sequential batches of 100), and read the billing tags of up to 100 sessions at once (‘get_session_tags’). Only available in private mode (via devin.ai endpoints).Input Parameters
string ("list" | "get" | "create" | "assign_session" | "assign_sessions" | "get_session_tags")
required
Operation to perform: ‘list’, ‘get’, ‘create’, ‘assign_session’, ‘assign_sessions’, or ‘get_session_tags’.
ID of the Devin session to assign. Required for ‘assign_session’. Assigning replaces the session’s current billing tag.
Map of Devin session ID -> billing tag ID (1-500 entries). Required for ‘assign_sessions’. A null value marks the session as miscellaneous (no billing tag). Assignments are sent to the API in sequential batches of 100. Partial success: sessions or tags not found in the organization, or sessions you can’t write, are reported per entry.
Session ids to look up billing tags for, at most 100. Required for ‘get_session_tags’. Both ‘devin-abc123’ and ‘abc123’ are accepted. Each session comes back with its billing tag, as having no billing tag when it was categorized as Miscellaneous or its tag was deleted, as never categorized, or as not found.
Billing tag id. Required for ‘get’. For ‘assign_session’, the tag to assign; omit it to mark the session as miscellaneous (no tag).
Output
Devin blueprint test
Devin blueprint test
Slug:
DEVIN_MCP_DEVIN_BLUEPRINT_TESTTest a candidate blueprint YAML in an authoring VM: start → run initialize → fix → run initialize again (clean VM) → run maintenance → then persist with update_environment_config.Input Parameters
Operation: start, run, get, cancel, or close.
Platform labels matching YAML “runs-on” values; use “default” for the default platform. Defaults to [platform].
Maximum time to wait for a run. Capped below the 15-minute approval window; use
get to keep following a longer run.Output
Devin code scan manage
Devin code scan manage
Slug:
DEVIN_MCP_DEVIN_CODE_SCAN_MANAGEManage Devin code scans (sometimes referred to as “security scans” or “Devin Security Swarm”). Targets the authenticated organization and requires the org code scan view permission for read actions or use permission for creation and remediation. Code scanning must be enabled for the org.The “create_profile” action creates an org-owned scan profile that mutates the organization’s scan configuration. The profile’s mode is fixed at creation; the guidance fields define the scan’s objective for non-security scan types. The “update_profile” action edits an existing org-owned profile: only the fields you set are changed; omitted fields are left untouched. Pass an empty string to clear a guidance field. For a ‘security’ profile the edit is not applied directly: it is posted to this session as an approval card showing the current and proposed profile, and takes effect only once the user approves it there. When the result says the change is awaiting approval, tell the user to review the card and never claim the profile was updated; a newer suggestion for the same profile supersedes the pending one.A profile is a reusable configuration that codifies a team’s scanning strategy. For ‘security’ scans it is user-facing (the Security page shows profiles by name): list them so the user can pick one, create a new one, or run with the defaults. For every other scan type it is an implementation detail: do not say “profile” to the user, list profiles for them, or ask them to pick or name one. Infer what they want from the conversation and submitted setup form, then create or reuse a profile yourself to match. Let the form explain its settings instead of repeating them in prose. When building the configuration:- Establish the scan type FIRST: call “list_scan_types” to see which types the org may use. The full set is ‘security’, ‘performance’, ‘db-queries’, ‘test-coverage’, ‘dead-code’, ‘code-quality’, ‘cleanup’ (behavior-preserving cleanup of messy, redundant, over-built code), ‘telemetry’, ‘accessibility’, ‘compliance’, ‘migration-docs’, and ‘general’ (no fixed domain — the profile defines the objective), but orgs without general scans enabled can only use ‘security’; never use a type that is not listed. Infer the type from what the user asked for instead of presenting the types as a menu: a standard type only when they asked for its whole domain (its name or an obvious synonym); anything narrower, merely adjacent, or uncovered (“camelCase variable names”, “memory leaks”) is a ‘general’ scan aimed at exactly that — never round a request to the nearest standard type. State the choice in a clause so they can correct it, and only ask when the request is genuinely ambiguous between two standard types. A profile only appears in scan creation for its own type, so always set scan_type explicitly to that type; a profile created with the wrong type will not show up where the user expects it.
- Infer focus and exclusions from the conversation and optional form guidance: problems or areas to focus on, parts of the codebase to skip (e.g. generated code, vendored deps, test fixtures), and anything that is always critical or never worth reporting. Map the answers onto fields yourself — threat_model_guidance (what to look for; the UI calls this the “scan model”), include/exclude globs (which files are scanned), triage_guidance (dedup/priority policy) — and never ask the user to fill in fields or enumerate them. Blank guidance uses defaults while preserving earlier instructions; do not ask again. Only set investigation, validation, report, or remediation guidance if the user volunteered that detail (validation guidance says how to build and run the code to confirm findings; without it that phase is skipped).
- Default non-security discover scans to a Slack DM summary to the requester when the scan finishes. Do not ask about communication or change it unless the user explicitly requests an override; store it as communication_guidance. Profiles are shared across the org: when reusing one that has no communication_guidance, fill it in with “update_profile” (create an org-owned copy for an account-owned profile that cannot be edited); if its guidance differs from what the user wants, create a new profile rather than overwriting it. Security and ingest scans do not run this phase.
- Reuse an existing profile (“list_profiles” with scan_type set to the chosen type, then “get_profile”) only when its description clearly covers what the user asked for; otherwise create a new one. A profile for a different scan type is never an option for this scan — do not propose, reuse, or retype one. If you reuse one, say so in the user’s terms (“your team has scanned for this before — I’ll reuse that setup”).
- If the user’s findings already come from their own scanner or security process, create an ingest-mode profile (mode=‘ingest’) instead of a discover profile: ingestion_source_guidance is then required in practice and must be concrete (where findings live, how to authenticate — reference credentials by org secret name, never inline a token), with optional post_ingestion_guidance for triage policy. Ingest profiles have no scope globs or scan model.
Input Parameters
Profile mode: ‘discover’ (default) or ‘ingest’ (‘create_profile’ only; the mode is fixed at creation).
Full names of the repositories to scan together as one scan (‘create_scan’; provide exactly one of repo_name or repos). The first entry is the scan’s primary repo. Use this whenever the user wants more than one repository covered.
string ("list_scans" | "list_findings" | "list_scan_types" | "list_profiles" | "get_profile" | "create_profile" | "update_profile" | "create_scan" | "update_scan" | "scan_new_commits" | "get_session" | "remediate_finding" | "remediate_findings")
required
The operation to perform: ‘list_scans’, ‘list_findings’, ‘list_scan_types’, ‘list_profiles’, ‘get_profile’, ‘create_profile’, ‘update_profile’, ‘create_scan’, ‘update_scan’, ‘scan_new_commits’, ‘get_session’, ‘remediate_finding’, or ‘remediate_findings’. ‘list_scan_types’ returns the scan types enabled for this org — only ever use those types (when general scans are disabled, only ‘security’). Pick the type from the user’s request yourself rather than presenting the types as a menu; use ‘general’ for anything no standard type covers. ‘create_scan’ launches the scan: only call it after the user has explicitly confirmed the setup by submitting code_scan_setup_form, or confirming a text summary when the form cannot be used. Preserve approved repositories, objective, effort, and notifications. ‘scan_new_commits’ re-runs an existing, completed scan against only the commits landed since it last ran — use it (not ‘create_scan’) whenever the user wants an existing scan brought up to date, e.g. “scan for new commits” or “rescan”; findings accumulate on the same scan. ‘update_scan’ changes the profile an existing scan applies to its future incremental runs (manual scan_new_commits and scan_new_commits automations alike) — use it (not a new scan) when the user wants an existing scan moved onto a different or revised profile. ‘get_session’ retrieves the latest run’s orchestrating session link for a scan.
Scan effort (‘create_scan’ only): ‘normal’ (default) uses lower model reasoning effort and larger investigation batches — faster; ‘deep’ runs the full pipeline, tracing each finding further and validating findings one at a time — most thorough, slower. Only security scans offer the choice; every other scan type uses ‘normal’. Explain the trade-off in terms of time and thoroughness only (never cost or usage).
Filter findings to this scan (‘list_findings’), or the scan ID containing the finding(s) (‘remediate_finding’, ‘remediate_findings’), or the scan to re-run (‘scan_new_commits’), or the scan to reconfigure (‘update_scan’), or the scan whose orchestrating session to retrieve (‘get_session’).
One-sentence, user-facing description of the proposed change, shown on the approval card when ‘update_profile’ targets a ‘security’ profile.
Filter findings to these severities: ‘critical’, ‘high’, ‘medium’, ‘low’ (‘list_findings’ only).
Filter to this repository (‘list_scans’ and ‘list_findings’), or the full repository name to scan (‘create_scan’; provide exactly one of repo_name or repos).
Type of scan, e.g. ‘security’ or ‘general’ (‘create_scan’, ‘create_profile’, ‘list_profiles’). For ‘list_profiles’: only return profiles for this scan type — always pass the chosen type so profiles for other types are never shown or considered. For ‘create_scan’: must match the chosen profile’s scan type; defaults to the profile’s type, or ‘security’ when neither a profile nor a type is given. Every non-security scan type requires a profile_id — the profile defines the scan’s objective. For ‘create_profile’: the type of scan the profile configures (defaults to ‘security’). For ‘update_profile’: the new scan type of the profile.
The profile ID. Required for ‘get_profile’, ‘update_profile’ and ‘update_scan’ (the profile future incremental runs use; must match the scan’s type and mode); optional for ‘create_scan’.
Repository-relative paths to restrict scans to (‘create_profile’ and ‘update_profile’). Include globs do not widen an existing restriction; pass [] with ‘update_profile’ to lift it.
Findings of scan_id to fix together (‘remediate_findings’; at least one). One session plans and groups them, then opens one pull request per group.
Interactive mode (‘create_scan’, security scans only): true pauses the scan once it has drafted its threat model and opens a session where the user reviews and adjusts it before investigation starts; false (default) runs end to end unattended. For security scans explain the difference, then ask. Never offer or set it for any other scan type — those always run unattended.
Finding severities runtime validation covers, from ‘critical’, ‘high’, ‘medium’, ‘low’ (‘create_profile’ and ‘update_profile’; security profiles only). Omit for the default (critical, high, medium); pass [] with ‘update_profile’ to restore it.
Guidance for the communication phase: what happens when the scan finishes, e.g. Slack the requester a summary, or ping the code owners of each finding with it and its fix (‘create_profile’ and ‘update_profile’). Always ask the user about this for non-security discover profiles. Security and ingest scans do not run this phase.
Output
Devin find setting
Devin find setting
Devin knowledge manage
Devin knowledge manage
Slug:
DEVIN_MCP_DEVIN_KNOWLEDGE_MANAGEManage Devin knowledge notes and suggestions via a single action-based tool.Input Parameters
Deprecated alias for
content. Prefer content; this is accepted only for backward compatibility with older callers/skills.string ("list" | "get" | "create" | "update" | "delete" | "folders" | "list_suggestions" | "view_suggestion" | "dismiss_suggestions")
required
Operation to perform. Notes: ‘list’, ‘get’, ‘create’, ‘update’, ‘delete’, ‘folders’ (returns the folder structure with per-folder note counts). Suggestions: ‘list_suggestions’, ‘view_suggestion’, ‘dismiss_suggestions’.
Search query for ‘list’ — case-insensitive substring match across note name, trigger, and content. Useful for dedup and conflict detection.
Filter ‘list’ results to notes in a specific folder (e.g. ‘Dana/SubFolder’ or ’/’ for root notes only).
Pin this note to a specific repo (owner/repo format). For create/update: sets the pinned repo. For list: filters results to notes pinned to this repo.
Output
Devin list integrations
Devin list integrations
Slug:
DEVIN_MCP_DEVIN_LIST_INTEGRATIONSList native integrations and MCP servers for the organization, with status and settings/install URLs.Returns a JSON object with two arrays: “integrations” (native integrations like GitHub, Jira, Slack) and “mcp_servers” (marketplace MCP servers). Each entry includes whether it’s installed and a path-only URL to the settings/setup page (no host, works with custom enterprise domains). Pass query to look up a specific server by name — an empty “mcp_servers” array then means it is not in the marketplace (install it as a custom server with devin_mcp_server_manage instead).Input Parameters
Case-insensitive substring matched against name, slug, and description. Use it whenever you are looking for one specific integration or MCP server (e.g. query=‘whimsical’); the unfiltered catalog is large.
Filter results — ‘all’ (default), ‘installed’, or ‘not_installed’.
Output
Devin mcp server manage
Devin mcp server manage
Slug:
DEVIN_MCP_DEVIN_MCP_SERVER_MANAGEManage MCP server installations for the organization: install (from the marketplace or custom), update, delete, enable, or disable. Marketplace installs use the one-call fast path. Before a custom install, use web research to confirm the official endpoint and authentication method. For an API key, use request_secret and pass only auth_header_secret_name. Never ask the user to paste a secret into normal session text. The pre-approval check rejects OAuth when automatic registration cannot complete. Install, update, and delete show the approval card unless the call is pre-approved, so do not send another approval request. Only available in private mode.Input Parameters
Server URL. Required for HTTP/SSE transport on a custom ‘install’. On a marketplace ‘install’, only to select a URL variant (e.g. a region): it must keep the scheme and host of one of the manifest’s URLs; omit it for the default.
Operation to perform. ‘resolve_install’ is read-only: it resolves a marketplace server or validates a custom server before the approval card.
‘none’, ‘auth_header’, or ‘oauth’. Before a custom install, research the provider’s official documentation and select a supported method. Never put a credential value in this tool call.
Installation id. Required for ‘update’, ‘delete’, ‘enable’, and ‘disable’. Find it with devin_list_integrations.
Pre-registered OAuth client ID for a custom install. Client IDs are identifiers, not secret values. If the provider also requires a client secret, configure that secret in Settings instead of putting it in session text.
How to format auth_header_secret_name: ‘bearer’ creates ‘Bearer ${SECRET_NAME}’; ‘raw’ creates ’${SECRET_NAME}’. Defaults to ‘bearer’.
Optional override: ‘org’ (Organization) shares one credential org-wide; ‘user’ (Personal) has each member connect their own account. Omit unless the user requests a scope; the marketplace default (or ‘org’) applies.
Protected Brain secret name for a custom auth_header install. First use request_secret and tell the user never to paste the value into normal session text. Pass only the name here; the credential value is never included in the install card. The name must start with MCP_ to limit access to MCP-specific secrets. For an org installation, use an org-scoped saved secret. A temporary secret works only in the current session.
For ‘install’: the marketplace server’s slug or exact name (e.g. ‘notion’ or ‘Notion’); no prior devin_list_integrations call is needed. Transport, command, and auth come from the marketplace manifest and cannot be overridden; url may only select a variant the manifest allows. Omit to install a custom server.
Output
Devin oncall manage
Devin oncall manage
Slug:
DEVIN_MCP_DEVIN_ONCALL_MANAGEManage Devin Oncall. Oncall resources mirror automations (same triggers / tools / limits / session_settings / session / run_as / metadata groups and the same merge-patch update semantics) — only the action is implicit: a responder triages one Slack channel or the account’s PagerDuty incidents (posting its findings as incident notes), an incident investigates one incident channel, a report summarizes responders/incidents on a schedule.Resources and actions:- responder: list, get, create, validate_create, update, validate_update, delete, issues (a responder’s open issues, paged via first/after/order_by).
- incident_settings: get, update, validate_update (one settings object per org; PATCH).
- incident: list (report_id= returns that report’s next-run period, candidates and period counts), get (Slack channel, parent session, report ids — the incident report is the ‘incident-report.md’ attachment on the parent session; fetch it with devin_session_interact get_attachments), stop, resume.
- report: list, get (config + live membership), create, validate_create, update, validate_update, delete, runs (GET history, or POST a new immutable run when incident_ids is supplied), run (one run with its frozen sections and counts).
- schemas (no resource): default_net_policy, the governing security profile (name, whether it governs the network, MCP servers it allows), default_devin_mode / available modes, default_runbook, slack_connected, pagerduty_connection_id (non-null iff PagerDuty responders can be created; pagerduty:* trigger schemas are listed only then). ingest_dashboard (no resource): queue an unattended run that turns a Datadog/New Relic dashboard into Oncall knowledge.
Input Parameters
Responder / incident_settings tool grants for spawned sessions: mcp_servers (only telemetry servers the org has installed — check devin_list_integrations), slack_channels, linear_enabled. Omit slack_channels on responder create to use the default: watched channels for channel-scoped triggers, All public channels for channel-less triggers. Explicit permissions override the default; matching a message never adds its channel to Slack tool permissions. Update merges per member; lists replace wholesale.
string ("list" | "get" | "create" | "validate_create" | "update" | "validate_update" | "delete" | "issues" | "stop" | "resume" | "runs" | "run" | "schemas" | "ingest_dashboard" | "report_context" | "responder_issues" | "report_stats" | "digest_runs" | "incident_candidates" | "submit_report_run")
required
Operation to perform on ‘resource’: ‘list’, ‘get’, ‘create’, ‘validate_create’, ‘update’, ‘validate_update’, ‘delete’, ‘issues’ (responder), ‘stop’/‘resume’ (incident), ‘runs’/‘run’ (report). Without a resource: ‘schemas’ (defaults, available modes, Slack connectivity) or ‘ingest_dashboard’. The flat legacy names (report_context, responder_issues, …) are aliases for older prompts; do not use them in new ones.
Responder limits: max_acu_limit per triage session and invocations rate (server default 150 per hour). Omit unless the user asks.
Identity spawned sessions run as: {‘type’: ‘organization’} (default; the org system user, recommended so triage survives the creator leaving) or {‘type’: ‘creator’}.
Responder: whether it triages new messages (default true). incident_settings: whether the incident feature is on for the org.
Responder investigation instructions. Omit by default — the server applies ‘Investigate the root cause to the best of your abilities.’ Only supply one when the user gives concrete investigation steps; pass ” on update to reset to the default.
Responder / incident_settings / report session tags (cost codes) spawned sessions bill to, validated as the run identity. Orgs that require session tags must supply exactly one allowed tag; otherwise omit.
Responder labels. Recommend {‘team’: …} always and {‘service’: …} when obvious from the channel/runbook; reports can select members by these tags (responder_tags). Update merges per key; a null value deletes the key.
Issue ordering for responder ‘issues’: ‘last_activity’ (default) or ‘occurrence_count’. A cursor is only valid for the ordering it was minted under.
What to operate on: ‘responder’ (a triage bot for one Slack channel or for PagerDuty incidents), ‘incident_settings’ (the org’s single incident-channel config), ‘incident’ (an incident channel + investigation session), or ‘report’ (recurring report over responders/incidents; runs live under it). Omit only for ‘schemas’ and ‘ingest_dashboard’.
Responder create (required)/update. Up to 10 triggers from one provider, OR’d, each with its own conditions; no two identical. Slack: ‘slack:message’ and/or ‘slack:usergroup_mentioned’ triggers, all in one Slack workspace. Usergroup mentioned requires a usergroup eq filter with a real Slack usergroup ID (S…), not a handle or user ID, in every OR group. Keep a channel restriction by default; only omit it when the user intends to watch across channels, explain the broader scope, and include at least one completed filter in every OR group (is_thread_reply alone does not count). Channel-scoped example: [{‘event_type’: ‘slack:message’, ‘conditions’: {‘any’: [{‘all’: [{‘field’: ‘channel’, ‘operator’: ‘eq’, ‘value’: ‘C0123ABC456’}]}]}}]. Ask the user for the channel ID (or find it via devin_list_integrations / Slack tools); never invent one. PagerDuty responder: ‘pagerduty:incident_*’ triggers (default just ‘pagerduty:incident_triggered’); an event type may repeat with different conditions, plus pagerduty_connection_id. Never mix the two providers. On update the list replaces wholesale.
Report id for report get/update/validate_update/delete/runs/run, and the optional filter for incident list (returns that report’s candidate period + incidents to classify).
Report ‘runs’ POST: incident ids classified into the report for this period (from incident list with report_id). Supplying this (even []) submits a run instead of listing runs.
Report create only: {‘slack’: {‘channel_id’: ‘C…’}} delivers each run to that channel. Recommend it when the user names a channel or one is obvious (e.g. the team’s channel); never invent an ID. Cannot be changed on update.
incident_settings: response mode, ‘proactive’ or ‘one_shot_rca’. Omit to preserve it; pass ” to clear it to the application default.
incident_settings: Slack workspace ID incidents are watched in (omit for one workspace). Pass ” on update to clear the stored workspace.
Report create/update: tag-managed membership — every responder whose metadata matches ALL these key/values (e.g. {‘team’: ‘payments’}), resolved live at each run. Preferred when the user describes members by team/service.
Report create/update: natural-language description of which incidents belong in this report (e.g. ‘incidents affecting the payments service’). Omit for a responders-only report; pass ” on update to clear it.
Responder: devin_mode (omit on create for the server default — the best of ultra > fusion > normal the org can use; Ultra is recommended for Oncall) and net_policy (omit on create for the default from ‘schemas’: the deployment Git Manager-only policy, or the governing security profile’s allowlist when one governs the network; explicit null = unrestricted, only when the user asks). On update, preserve unless asked.
incident_settings: whether Devin auto-joins new public channels matching the prefix (default true — keep it on unless the user wants to invite Devin manually).
incident_settings: prefix for auto-created incident channels (server default ‘inc-’). Pass ” on update to clear the stored prefix.
PagerDuty responders only: the account’s PagerDuty connection id the responder posts incident notes through — read it from ‘schemas’ (pagerduty_connection_id; null means PagerDuty is not connected / not enabled for this org, so no PagerDuty responder can be created). Required with pagerduty:* triggers, rejected with slack:message.
Report create/update: fixed responder membership (ids from responder list). Mutually exclusive with responder_tags; on update the list replaces wholesale.
incident_settings: devin_mode plus the staged investigation overrides initial_investigation_mode and thread_session_mode. Omit either override to use the application default; explicit null clears it. net_policy follows the deployment default and explicit-null semantics. The legacy session_settings alias is also accepted.
Output
Devin playbook manage
Devin playbook manage
Slug:
DEVIN_MCP_DEVIN_PLAYBOOK_MANAGEManage Devin playbooks with a single action-based tool.Input Parameters
Deprecated alias for
content. Prefer content; this is accepted only for backward compatibility with older callers/skills.Automation macro identifier. Must start with ’!’ followed by word characters (letters, digits, underscores), e.g. ‘!my_macro’. Optional.
Operation to perform: ‘list’, ‘get’, ‘create’, ‘update’, or ‘delete’.
JSON Schema (Draft 7) defining the required output contract for sessions using this playbook. Max 64KB. Optional for ‘create’ and ‘update’.
Output
Devin review manage
Devin review manage
Slug:
DEVIN_MCP_DEVIN_REVIEW_MANAGETrigger a Devin Review for a pull/merge request, fetch the latest review status for one, or fetch a completed review’s findings. Only available in private mode (via devin.ai endpoints). The repo must be connected to the authenticated organization.Reviews run asynchronously: ‘trigger’ accepts the review (status starts as ‘pending’); poll with ‘get_status’ until it reaches ‘completed’, then use ‘get_findings’ to read the findings (‘get_findings’ returns an error until the review completes).Input Parameters
‘trigger’ starts a Devin Review for the PR; ‘get_status’ fetches the latest review status for the PR; ‘get_findings’ fetches the findings of a completed review. By default, only the severities the org delivers to sessions are returned; the rest remain available on the Devin Review page and can be fetched by passing the severity explicitly. ‘report_triage’ reports what you did about each finding of a review in a single call; findings marked needs_human are then posted on the PR.
Full URL of the pull/merge request (e.g. ‘https://github.com/owner/repo/pull/123’).
get_findings only: filter findings to one severity (‘bug’, ‘security’, ‘flag’, or ‘info’). Bug means a concrete correctness, regression, or repo-rule violation; security means a vulnerability or hardening gap; flag means something needing further investigation that may or may not be a bug; info means an informational note. Omit to return only the severities the org delivers to sessions; the remaining severities stay available on the Devin Review page and can be fetched by passing the severity explicitly.
get_status/get_findings: commit SHA to look up (full or short prefix). Defaults to the PR’s current head commit. report_triage: required; the commit the findings were reported for.
Output
Devin schedule manage
Devin schedule manage
Slug:
DEVIN_MCP_DEVIN_SCHEDULE_MANAGEManage scheduled Devin sessions (deprecated in favor of Automations).Note: “create” is only for organizations not yet migrated to Automations. For migrated organizations the API rejects it with 403 — schedule work by creating an automation with a schedule trigger via ‘devin_automation_manage’ instead. The other actions keep working for existing schedules.Input Parameters
The operation to perform: ‘list’, ‘get’, ‘create’, ‘update’, or ‘delete’.
VM platform for sessions spawned by this schedule (e.g. ‘windows’). When omitted, sessions fall back to the org’s default platform at trigger time. Must match a platform configured for the organization (case-insensitive); unknown values are rejected with a 400 that lists the available platform labels.
Devin session ID to send the prompt to instead of starting a new session. Only valid for one_time schedules. Use this to schedule a message to be sent to an existing session at a future time (e.g. to wake yourself up later).
Output
Devin session create
Devin session create
Slug:
DEVIN_MCP_DEVIN_SESSION_CREATECreate one or more child Devin sessions via the v3 REST API.Only use this when prompted to; do not use it to parallelize your own work by default.Before calling this, invoke the managing-child-sessions skill for guidance.The returned session_id values do NOT include the “devin-” prefix. To use them with devin_session_interact, prepend “devin-” (e.g. session_id “abc123” becomes “devin-abc123”).Input Parameters
Tags to apply to ALL created sessions (e.g. a shared run or campaign identifier). Each session spec can also include a ‘tags’ key to override this value for that session.
List of repository identifiers in ‘owner/repo’ format (e.g. [‘myorg/myrepo’]) to restrict ALL created sessions to. Each session spec can also include a ‘repos’ key to override this value for that session.
Prefer leaving this unset. When omitted (or set to ‘inherit’), the child inherits the parent session’s placement — its VM platform AND its outpost (BYOB) pool — so a Windows session spawns Windows children, and a session running on an outpost pool spawns children on the same pool, by default. Only set this when the user explicitly asks for a different machine type or a specific outpost pool: pass a platform label (e.g. ‘linux’, ‘windows’) or the name of an outpost pool. Each session spec can also include a ‘platform’ key to override this value for that session. Any value must match either a platform label or an outpost pool name registered for the org (built-in platforms take priority when a name matches both).
Devin agent mode for ALL created sessions, by the id of an option in the user’s agent picker. Picker names → ids: ‘Normal’ = ‘normal’, ‘Fast’ = ‘fast’, ‘Fusion’ = ‘fusion’, ‘Lite’ = ‘lite’, ‘Ultra’ = ‘ultra’, ‘SWE-2 Medium’ = ‘swe-2-medium’, ‘SWE-2 High’ = ‘swe-2-high’, ‘SWE-2 Max’ = ‘swe-2-max’. Always refer to modes by picker name when talking to the user, never by id. Batch-level only — a ‘devin_mode’ key inside a session spec is rejected with guidance to move it to this batch-level field. When omitted, children run on the parent session’s agent. Gated like the picker (feature flags and enterprise agent-mode policies); a 400 lists the modes available to this user.
Devin Boards ticket ids (from devin_ticket_manage) that ALL created sessions are dispatched to work on. Each session spec can also include a ‘board_ticket_ids’ key to override this value for that session. The server validates the ids against the org, persists them on the child and tags it ‘board’; the CHILD claims the tickets itself on its first turn, so do not claim them before dispatching.
JSON Schema (Draft 7) applied to ALL sessions that don’t specify their own. Only pass this when you need a task-specific result from the child; without it no structured output is requested. The child reports output conforming to the schema via provide_structured_output; poll it via devin_session_interact(action=‘get’) — the structured_output field is populated without interrupting the child.
Output
Devin session events
Devin session events
Slug:
DEVIN_MCP_DEVIN_SESSION_EVENTSInspect events within a Devin session — list summaries, fetch full details, or search.Actions:- list: Paginated event summaries with optional filters.
- details: Batch-fetch full event contents by event_id or by offset+limit range.
- search: Full-text search across event contents.
Input Parameters
0-based offset into the session timeline for details (details only, mutually exclusive with event_ids). Use with limit.
Filter by category — shell, file, search, browser, mcp, git, message, status, secret, todo, recording, knowledge, playbook, webhook, lifecycle, other (list/search only, mutually exclusive with event_types).
Output
Devin session gather
Devin session gather
Slug:
DEVIN_MCP_DEVIN_SESSION_GATHERWait for multiple Devin sessions to reach a settled state before returning.A session is “settled” when it is no longer actively working:- status is “exit” (finished or terminated)
- status is “error” (crashed)
- status is “suspended” (sleeping)
- status is “running” with status_detail “finished”, “waiting_for_user”, or “waiting_for_approval”
Input Parameters
Session IDs to wait on (must include the ‘devin-’ prefix, e.g. ‘devin-abc123…’). Maximum 50.
Ceiling on the interval between status checks. 0 (default) auto-scales with the number of sessions (5-30s); polling starts at 2s and ramps up to the ceiling.
Output
Devin session interact
Devin session interact
Slug:
DEVIN_MCP_DEVIN_SESSION_INTERACTConsolidated tool for all interactions with a single Devin session.Actions, and what each returns: get: the session’s details — session_id, title, status, status_detail, is_archived, url, acus_consumed, tags, child_session_ids (the sessions it spawned; this is how you look up a session’s, or your own, children — they come back without the “devin-” prefix, which these tools still accept), pull_requests, structured_output. Fields are omitted when empty. message: sends a message (resumes the session if it is sleeping), and returns the same session details as “get”. sleep / terminate / archive / unarchive: performs the lifecycle action and returns the resulting session details, same shape as “get”. Archiving cascades to child sessions; unarchiving does not. get_messages: the session’s message history — per message, its source, created_at, event_id and body (long bodies are truncated; fetch a specific message in full via devin_session_events action=“details” with its event_id). Paginated via first / after. get_attachments: the session’s attachments — name, content_type, url. Not paginated. set_tags: replaces the session’s tags and returns the resulting tag list.Input Parameters
Tag list (required when action is ‘set_tags’). Replaces all existing tags on the session — this is a full replacement, not a union/append.
Page size for get_messages pagination (v3 cursor param). get_attachments is not paginated — v3 returns the full list.
string ("get" | "message" | "sleep" | "terminate" | "archive" | "unarchive" | "get_messages" | "get_attachments" | "set_tags")
required
Operation to perform — get, message, sleep, terminate, archive, unarchive, get_messages, get_attachments, set_tags.
Target Devin session ID, e.g. ‘devin-abc123…’ (the ‘devin-’ prefix is optional — bare IDs, such as the ones in child_session_ids, also work).
Optional list of already-uploaded attachment URLs to attach to the message (only used when action is ‘message’). Each URL must belong to the session’s org — same mechanism as the initial prompt’s attachments.
Only used when action is ‘message’. When true, you are notified (and woken if sleeping) once when this session next settles (finishes, blocks, or exits). One-shot: consumed on delivery — pass it again on a later message to re-subscribe. Works for any session you are allowed to message (descendants, same-automation siblings, and sessions currently claiming tickets on a board you manage).
Output
Devin session search
Devin session search
Slug:
DEVIN_MCP_DEVIN_SESSION_SEARCHSearch and filter Devin sessions. All filters are optional and combined with AND.The returned session_id values do NOT include the “devin-” prefix. To use them with devin_session_interact, prepend “devin-” (e.g. session_id “abc123” becomes “devin-abc123”).Input Parameters
Filter by origin. Valid values: ‘webapp’, ‘slack’, ‘teams’, ‘api’, ‘linear’, ‘jira’, ‘scheduled’, ‘cli’, ‘other’.
Filter by billing tag. Values are billing tag ids (see devin_billing_tag_manage) or the special values ‘miscellaneous’ (explicitly tagged Miscellaneous) and ‘uncategorized’ (never tagged); a session matches if it matches any value. Only available for organizations with billing tags enabled — the search fails with an error otherwise.
Output
Devin user list
Devin user list
Slug:
DEVIN_MCP_DEVIN_USER_LISTList the current organization’s users, one membership kind per call. Rows are user_id | email | name (roles appended with include_roles).Input Parameters
‘direct’: users added to the organization individually, with roles assigned to them. ‘idp’: users who are members via an identity-provider (SSO/SCIM) group mapped to the organization, with roles inherited from the group. A user can be both; list both to enumerate every member.
Output
Generate wiki
Generate wiki
List wiki repos
List wiki repos
Read wiki contents
Read wiki contents