Tool Catalog (generated)

Machine-generated signature index for the Leif MCP tools, built from the live FastMCP tool schemas. The exhaustive companion to the curated per-namespace reference pages.

This page is generated from the live FastMCP tool schemas — the parameters and defaults are exactly what the server accepts. It is the exhaustive, never-stale signature index; the curated namespace pages remain the readable guides with prose, gotchas, and worked examples.

cwa_* — ConnectWise Automate (RMM)

Find managed computers, run scripts/commands, read history. Note: only saved-script execution (cwa_execute / cwa_run_powershell) returns real stdout; cwa_run_command type 2 always returns Output: “ERR”.

Curated guide: ConnectWise Automate (RMM).

cwa_await_callback

Wait for a pending CWA PowerShell callback result. Use this after cwa_run_powershell returns status: Timeout with a callback_token. The shim script was successfully queued — it just hasn’t run yet because the CWA agent scheduler tick (~80-130s) hadn’t fired within the first call’s poll window. This tool polls the local SQLite store (cheap — no CWA API calls) for the token until the managed PC POSTs its output or the token expires. Args: token: The callback_token from a previou…

cwa_execute

Run a script on a computer.

cwa_get

Get or list ConnectWise Automate records.

cwa_probe_command_endpoint

Probe candidate CWA command endpoint paths to find the working one. Tries POST against all known path variants and reports HTTP status codes. 200/202 = working; 405 = path exists, wrong method; 404 = not found. Also caches the winner so subsequent cwa_run_command calls use it. Args: computer_id: A valid computer ID to use in path substitution.

cwa_raw_request

Make a raw authenticated CWA API request and return the full response. Unlike other tools, this never raises on 4xx/5xx — it returns the raw status_code, headers, and body_text so you can diagnose schema mismatches or explore endpoints. Args: method: HTTP method (GET, POST, PUT, DELETE). path: API path beginning with / (e.g. /Commands). body: Optional JSON body for the request. params: Optional query parameters.

cwa_run_command

Queue a command on a CWA-managed computer. Verified live against POST /Computers/{id}/CommandExecute. IMPORTANT — output capture is limited by CWA itself: - command_type="2" (“Execute Command”) is fire-and-forget. The agent ALWAYS returns Output: "ERR" regardless of the actual stdout. Status reflects launch success only. To get real stdout, run a saved CWA script via cwa_execute / cwa_run_powershell (which use the /Batch/ScriptExecute “run now” endpoint and read output…

cwa_run_powershell

Run an ad-hoc PowerShell script via the CWA shim and return stdout. This is the only path in CWA that returns real script output. Command type 2 (“Execute Command”) always returns Output: "ERR" regardless of stdout — this tool exists to work around that. Callback mode (default when CWA_CALLBACK_BASE_URL is set): Instead of relying on CWA’s DiagnosticMessage field (which is truncated to ~80 chars), the user’s script is wrapped in a PowerShell snippet that captures all stdout/st…

cwa_script_history

Fetch recent script-run history for a CWA computer. Each record includes DiagnosticMessage — the full execution log with captured variable values (when the script has @ScriptEngineEnableLogger@ set to True). This is the primary stdout-capture surface in CWA. Important: The list endpoint truncates DiagnosticMessage to ~80 characters. To get complete output for a specific run, pass history_id (the history_id returned by cwa_run_powershell after a Timeout or compl…

cwa_search

Search ConnectWise Automate computers.

cwa_search_scripts

Search the CWA script library by name or OData condition. Use this BEFORE writing your own diagnostic logic — there is almost certainly already a “Get Eventlog Last 24h” or similar script in the library. Pass the discovered Id to cwa_execute to run it on a target computer; that is the only path that returns real stdout from CWA. Args: name: Substring matched as Name like '%<name>%'. condition: Raw condition string. Overrides name. page: 1-indexed page number. page_size: Items…

cwa_status

Get a quick ConnectWise Automate health overview.

hudu_* — Hudu (IT Documentation)

IT documentation system of record — companies, assets, articles, passwords, and their relationships. Entity-typed CRUD scoped by company_id; ‘delete’ is usually an archive via hudu_update (action=archive), and hudu_lookup bridges an external integration’s identifier to a Hudu resource and back.

Curated guide: Hudu (IT Documentation).

hudu_create

Create a Hudu record.

hudu_delete

Delete a Hudu record.

hudu_get

Get or list Hudu records by entity type.

hudu_lookup

Perform a special Hudu lookup.

hudu_search

Search across Hudu entities.

hudu_update

Update, archive, or unarchive a Hudu record.

repairshopr_* — RepairShopr

System of record for tickets, customers, invoices. Entity-typed CRUD plus comments and search.

Curated guide: RepairShopr.

repairshopr_comment

Add a comment to a RepairShopr ticket.

repairshopr_create

Create a RepairShopr record.

repairshopr_delete

Delete a RepairShopr record.

repairshopr_export

Slim paged export of tickets, customers, invoices, or POs for sync pipelines. Returns summary fields only, so responses stay far below the MCP output cap. Tickets filter server-side via since_updated_at; customers walk updated_at DESC pages and stop once records are older than since_updated_at (RS has no since filter for customers).

repairshopr_get

Get or list RepairShopr records.

repairshopr_search

Search across RepairShopr entities.

repairshopr_update

Update a RepairShopr record.

quickbooks_* — QuickBooks Online

Accounting. quickbooks_query is the SQL-like read workhorse; mutations use QBO’s SyncToken concurrency model. IDs are strings.

Curated guide: QuickBooks Online.

quickbooks_create

Create a QuickBooks record.

quickbooks_delete

Delete a QuickBooks record.

quickbooks_export

Export QuickBooks records to CSV with pagination.

quickbooks_get

Get a single QuickBooks record or list records.

quickbooks_invoice_class_detail

Return invoice class attribution for an exact QuickBooks invoice number. Exposes transaction- and line-level ClassRef values in a compact shape so callers can audit upstream location/class mappings without parsing the full QBO invoice payload.

quickbooks_query

Run a raw QuickBooks SQL-like query.

quickbooks_reconcile_invoice_classes

Bulk-audit expected class against actual QBO invoice line classes. Input rows require doc_number and expected_class; arbitrary caller context (for example RS invoice id, technician, or RS location) is echoed back. This tool is read-only and never changes QuickBooks.

quickbooks_report

Run a standard QuickBooks financial report.

quickbooks_update

Update a QuickBooks record.

cf_* — Cloudflare

Zones, DNS, SSL mode, settings, cache. IMPORTANT: most cf_* tools take their args inside a nested params object (shown below as the inner fields). DNS updates are full replacements.

Curated guide: Cloudflare.

cf_create_dns_record

Create a new DNS record in a Cloudflare zone. Args: params.zone_id: Cloudflare zone ID params.record_type: Record type (A, AAAA, CNAME, MX, TXT, SRV, etc.) params.name: Record name (e.g. ‘www’, ‘mail’, or ’@’ for root) params.content: Record value (IP, hostname, or text content) params.ttl: TTL in seconds; 1 = automatic (default) params.proxied: Route through Cloudflare proxy (default False) params.priority: Priority for MX/SRV records Returns: str: JSON with the created record details includ…

cf_delete_dns_record

Delete a DNS record from a Cloudflare zone. This is a destructive operation and cannot be undone. Use cf_list_dns_records to find the record_id before deleting. Args: params.zone_id: Cloudflare zone ID params.record_id: ID of the DNS record to delete Returns: str: JSON confirming deletion with the deleted record ID.

cf_get_ssl_mode

Get the SSL/TLS encryption mode for a Cloudflare zone. Modes: off, flexible, full, strict. ‘flexible’ is the most common cause of WordPress redirect loops. Args: params.zone_id: Cloudflare zone ID (get from cf_get_zone or cf_list_zones) Returns: str: JSON with the current ssl setting id, value, and editable flag.

cf_get_zone

Look up a Cloudflare zone by domain name and return its details including zone ID. Args: params.domain: The exact domain name to look up (e.g. ‘sanbornpropane.com’) Returns: str: JSON with zone id, name, status, plan, name_servers, and paused flag.

cf_get_zone_settings

Get all settings for a Cloudflare zone. Returns all configurable settings (SSL, security level, caching, minification, always_use_https, WAF, etc.) as a list of {id, value, editable} objects. Args: params.zone_id: Cloudflare zone ID Returns: str: JSON with list of all zone settings.

cf_list_dns_records

List DNS records for a Cloudflare zone. Args: params.zone_id: Cloudflare zone ID params.record_type: Optional filter by type (A, CNAME, MX, TXT, etc.) params.name: Optional filter by record name Returns: str: JSON with list of DNS records including id, type, name, content, ttl, proxied.

cf_list_zones

List all Cloudflare zones (domains) on the account. Returns a list of zones with id, name, status, name_servers, and plan. Use zone IDs from this list as input to other cf_* tools. Returns: str: JSON with success bool and result list of zone objects.

Parameters not in the seed catalog — see the namespace reference page.

cf_purge_cache

Purge cached assets for a Cloudflare zone. Either purge_everything=True (clears all cached files) or provide a list of specific file URLs to purge selectively. Args: params.zone_id: Cloudflare zone ID params.purge_everything: If True, purge all cached assets params.files: List of specific URLs to purge (alternative to purge_everything) Returns: str: JSON confirming the purge operation.

cf_set_ssl_mode

Set the SSL/TLS encryption mode for a Cloudflare zone. Valid modes: off - No SSL (HTTP only) flexible - SSL between visitor and Cloudflare; HTTP to origin full - SSL end-to-end; origin cert can be self-signed strict - SSL end-to-end; origin cert must be valid/trusted Use ‘full’ or ‘strict’ to fix WordPress/origin redirect loops caused by ‘flexible’. Args: params.zone_id: Cloudflare zone ID params.mode: One of: off, flexible, full, strict Returns: str: JSON with the updated ssl setting.

cf_set_zone_setting

Set a specific zone setting for a Cloudflare zone. Common setting IDs and example values: always_use_https -> “on” or “off” security_level -> “off” | “essentially_off” | “low” | “medium” | “high” | “under_attack” browser_cache_ttl -> integer (seconds) cache_level -> “aggressive” | “basic” | “simplified” development_mode -> “on” or “off” email_obfuscation -> “on” or “off” minify -> {“css”: “on”, “html”: “on”, “js”: “on”} rocket_loader -> “on” or “off” automatic_https_rewrites -> “on” or “off”…

cf_update_dns_record

Update (replace) an existing DNS record in a Cloudflare zone. All fields are required as this is a full replacement (PUT), not a partial update. Use cf_list_dns_records to find the record_id. Args: params.zone_id: Cloudflare zone ID params.record_id: ID of the DNS record to update params.record_type: Record type (A, AAAA, CNAME, MX, TXT, etc.) params.name: Record name params.content: New record value params.ttl: TTL in seconds; 1 = automatic params.proxied: Route through Cloudflare proxy para…

pve_* — Proxmox Virtual Environment

Node/storage/template inventory, LXC & VM lifecycle, in-guest exec, async task polling. Most tools take node + vmid; lifecycle ops are async and return a task UPID.

Curated guide: Proxmox Virtual Environment.

pve_lxc_create

Create an LXC container on a Proxmox node. Required inputs: node, vmid, hostname, ostemplate, storage, rootfs_size_gb, cores, memory_mb. Optional inputs: swap_mb, password, ssh_public_keys, net0, unprivileged, features, start. If the container will run Docker, set features=“nesting=1”. Without that, Docker-in-LXC will fail. Returns the async task UPID.

pve_lxc_destroy

Destroy an LXC container. This permanently deletes the container, so force=true is required.

pve_lxc_exec

Run a shell command inside an LXC using pct exec over SSH to the Proxmox host. Returns stdout, stderr, and return_code.

pve_lxc_list

List LXC containers with vmid, name, status, CPU, memory, and disk usage. If node is omitted, includes all nodes.

pve_lxc_shutdown

Gracefully shut down an LXC container and return the Proxmox task UPID.

pve_lxc_start

Start an LXC container and return the Proxmox task UPID.

pve_lxc_stop

Hard-stop an LXC container. This is abrupt power-off behavior, so force=true is required.

pve_node_status

Get CPU, memory, disk, uptime, and load data for each Proxmox node.

Parameters not in the seed catalog — see the namespace reference page.

pve_storage_list

List Proxmox storage pools with type, used, total, available, and content types. If node is omitted, lists storage on all nodes.

pve_task_status

Get the status of an async Proxmox task by node and UPID.

pve_template_list

List available LXC templates on a node and storage. Use this to find the exact Debian 12 template name before pve_lxc_create.

pve_vm_exec

Run a shell command inside a QEMU VM via the QEMU Guest Agent. Requires qemu-guest-agent installed and running inside the VM (and the VM’s config must have agent: 1). Returns stdout, stderr, and return_code. If the agent is unavailable, return_code will be non-zero and stderr will carry the host-side error message.

pve_vm_list

List QEMU VMs with vmid, name, status, CPU, memory, and disk usage. If node is omitted, includes all nodes.

growably_* — Growably / LeadConnector CRM

Contacts, opportunities, conversations, tags, messaging, workflows. Discover IDs via growably_get; use growably_manage_tags (not growably_update) for tags; confirm before send_message / trigger_workflow.

Curated guide: Growably / LeadConnector CRM.

growably_add_note

Add a note to a Growably contact.

growably_create_opportunity

Create a new opportunity in a Growably pipeline. Use growably_get entity_type=pipeline to discover pipeline/stage IDs. status: open, won, lost, abandoned (default: open)

growably_get

Get or list Growably/LeadConnector records. entity_type options: contact — list contacts or get by id opportunity — list opportunities or get by id pipeline — list all pipelines calendar — list calendars calendar_event — list events (filter: calendar_id, start_time, end_time) conversation — list conversations or get by id (filter: contact_id) conversation_messages — get messages for a conversation (requires: id) workflow — list workflows tag — list all tags custom_field — list custom fields l…

growably_manage_tags

Add or remove tags on a Growably contact. action: “add” or “remove” tags: List of tag strings Uses dedicated tag endpoints (not PUT contact) to avoid overwriting existing tags.

growably_send_message

Send a message to a Growably contact via SMS or Email. IMPORTANT: Always confirm the recipient, message content, and channel with Wayne before calling this tool. Once sent, messages cannot be recalled. type: SMS (default), Email, GMB, WhatsApp, IG, FB, Custom subject: Required when type=Email email_from: Optional sender address override for Email

growably_trigger_workflow

Trigger a Growably workflow on a contact. Use growably_get entity_type=workflow to discover workflow IDs. IMPORTANT: Confirm with Wayne before triggering workflows that send communications to customers.

growably_update

Update a Growably record. entity_type: contact, opportunity id: Record ID to update data: Fields to update (dict). For contacts, avoid updating tags directly here — use growably_manage_tags instead to prevent tag overwrites.

growably_upsert_contact

Create or update a Growably contact (matched by email/phone). If a contact with the same email or phone already exists it will be updated; otherwise a new contact is created. custom_fields format: [{“id”: “<field_id>”, “value”: ""}] Use growably_get entity_type=custom_field to discover field IDs.

gmail_* — Gmail (lowercase)

Search with Gmail query syntax, read, send, draft, attachments, labels. Use lowercase gmail_* (not capital-G Gmail:*). For customer-facing sends, use gmail_stage_send → gmail_confirm_send instead of sending directly.

Curated guide: Gmail (lowercase).

gmail_confirm_send

Confirm and deliver a staged email. Retrieves the staged send created by gmail_stage_send, sends it via Gmail, and removes it from the staging store.

gmail_create_draft

Create a Gmail draft. attachments is an optional list of absolute file paths on the Leif server, under the uploads folder (by default /home/superht/data/uploads), at most 10 files and 17 MB in total. A refused path creates no draft.

gmail_discard_staged_send

Discard a staged email without sending it.

gmail_get_attachment

Download a Gmail attachment.

gmail_get_labels

List Gmail labels.

Parameters not in the seed catalog — see the namespace reference page.

gmail_get_message

Get a single Gmail message.

gmail_get_recent_emails

Get recent Gmail messages with optional unread filtering.

gmail_list_staged_sends

List all pending staged email sends awaiting confirmation.

Parameters not in the seed catalog — see the namespace reference page.

gmail_mark_as_read

Mark a Gmail message as read.

gmail_search_emails

Search Gmail messages using query syntax.

gmail_search_emails_batch

Search Gmail messages with pagination until the requested limit.

gmail_send_email

Send a Gmail message. attachments is an optional list of absolute file paths on the Leif server. Files must be under the uploads folder (by default /home/superht/data/uploads, where gmail_get_attachment saves), at most 10 files and 17 MB in total. A refused path sends nothing.

gmail_stage_send

Stage an outbound email for review before it is sent. Use this instead of gmail_send_email for customer-facing messages. Returns a send_id. Call gmail_confirm_send with that id to actually deliver it, or gmail_discard_staged_send to cancel. Staged sends are stored locally and never leave the server until confirmed — the Gmail API is not contacted at this step. attachments is an optional list of absolute file paths on the Leif server, under the uploads folder (by default “…

sheets_* — Google Sheets

Read, write, and manage Google Sheets. sheets_read and sheets_write take a spreadsheet_id plus sheet/range; sheets_manage handles structural operations (add sheet, resize, format).

Curated guide: Google Sheets.

sheets_manage

Create spreadsheets, inspect metadata, or add sheets.

sheets_read

Read values from a spreadsheet.

sheets_write

Write, append, clear, or find/replace values in a spreadsheet.

add_* — add

add_email_filter

Create an email filter.

agent_* — agent

agent_dispatch

Start a scoped subagent run on the Leif host and return immediately. Edit agents (“edit”, “edit-probe”) are restricted to the principals in their allowed_callers. An “edit” run makes one change to the Leif repo in an ATC lane at an immutable base, checks every changed path against scope, runs the landing gate from main, and pushes atc/. It never lands or deploys. agent_result returns the manifest: base_sha, commit_sha, branch, changed_files, gate exit code and log tail. Args: agent:…

agent_list

List dispatchable Leif subagents with model ceilings, caps and cost hints. Call this before agent_dispatch to pick the right agent. Every agent runs read-only under its own scoped Leif principal; the model can only be lowered from the default, never raised above the ceiling.

Parameters not in the seed catalog — see the namespace reference page.

agent_result

Final report plus subtype, stop reason, cost, usage, note id, and status.

agent_runs

List recent dispatched runs, newest first (run_id, agent, status, cost, note_id). Use this to recover a lost run_id or to see what is queued/running before dispatching more. Compact rows only; call agent_result for a specific run’s report. Args: limit: Max rows (1-200, default 20). status: Optional exact filter: queued|running|done|failed|timeout|lost.

agent_status

Run status; cost, turns, and usage normally appear only after termination.

agent_wait

Wait server-side until a run is terminal or max_wait_s passes; one call, not a poll loop. Returns the run’s status either way, so dispatch + agent_wait replaces dispatch + N agent_status polls. Bounded: max_wait_s is capped (default cap 90s, under the MCP edge’s request timeout) and the call never blocks past it. If terminal is false, call agent_wait again. Args: run_id: The dispatched run to wait on. max_wait_s: Seconds to wait before returning the current status.

append_* — append

append_to_excel

Append rows to an existing Excel workbook.

apt_* — apt

apt_install

Install packages with APT on the local host.

apt_search

Search available APT packages on the local host.

backfill_* — backfill

backfill_event_log

Backfill historical conversation turns into the event log. source_format is claude_export, slack_canvas, or reconstruction_from_notes. raw_data_or_path is a file path, a JSON string, or (for slack_canvas) raw transcript text. Turns are appended with source="backfill" and a per-source confidence so reconstructed history is weighted differently from real-time turns. Optionally restrict to conversation_date_range [start_iso, end_iso].

bootstrap_* — bootstrap

bootstrap_context

Load scope-based living docs and the latest handoff for session bootstrap.

break_* — break

break_down_task

Create subtasks under a parent task.

calendar_* — Google Calendar

Curated guide: Google Calendar.

calendar_create_event

Create a calendar event.

calendar_delete_event

Delete a calendar event.

calendar_get_events

List events in a calendar.

calendar_get_free_busy

Get free/busy information for one or more calendars.

calendar_list_calendars

List all accessible calendars.

Parameters not in the seed catalog — see the namespace reference page.

calendar_update_event

Update a calendar event.

capture_* — capture

capture_action_items

Turn a conversation’s outcome into tracked work. Sort the conversation’s results into two buckets and submit once: - action_items: committed to-dos. Each becomes a task grouped under a single parent task named topic and linked to this conversation, so the whole list stays cohesive and traceable back to the chat. Each item is a dict with a required title and optional priority, category, due_date, due_time, description. - ideas: tentative thoughts not yet wor…

capture_deadline

Record a deadline or a deadline-less item to track. Pass due_date (ISO YYYY-MM-DD) for a hard deadline — it gets bucketed by the health check. Omit it to track important work that has no forced date; the stasis detector then watches it and surfaces it if it goes quiet for 3+ weeks. Capture high-confidence items only; if a date is ambiguous, confirm with Wayne first rather than guessing.

capture_task

Quick-capture a task from natural language with optional overrides. Severity: when priority is omitted, a P0/P1/P2 severity marker in the content can escalate the priority (issue #36, Finding 3) — P0/outage/incident → urgent, P1/blocker → high, P2 → medium — on top of the service’s existing urgency-keyword inference. An explicit priority always wins, and a marker never downgrades a stronger signal already detected. Category is intentionally not a parameter: server-side inf…

cleanup_* — cleanup

cleanup_old_checkpoints

Delete stale progress checkpoints.

commit_* — commit

commit_bill_entry_batch

Commit one exact financial batch after separate durable approval.

commit_check_deposit_batch

Commit one exact financial batch after separate durable approval.

complete_* — complete

complete_checkpoint

Mark a progress checkpoint complete.

complete_leif_task

Deprecated compatibility alias for complete_tracked_task.

complete_task

Deprecated compatibility alias for complete_checkpoint.

complete_tracked_task

Mark an actionable tracked task as done.

conversation_* — conversation

conversation_analytics

Analyze Leif conversation history across all stored sessions.

convert_* — convert

convert_excel_to_csv

Convert an Excel sheet to CSV.

council_* — council

council_result

Read this principal’s paged Council result; follow each seat next_offset. Optional seat: claude/codex/xai; offset >=0, limit 1-1000 characters. Default returns the first page of all seats, never an opaque output cap.

council_status

Read only the current authenticated principal’s Council receipt and status.

council_submit

Submit project text to all three advisory Council seats; return durable receipt. This sends artifact content to external model providers. authorization_scope carries the user’s existing explicit instruction and content-specific scope; it is attribution, never approval or a policy override. Respect platform review and sensitive-data restrictions. Do not ask again solely because this is a write. Retry identical requests with the same key; changed content fails. Fetch status and result; never re…

create_* — create

create_directory

Create a local directory.

create_file

Create or overwrite a local file.

create_project

Create a new project context.

create_task

Create a task with explicit fields. priority accepts None (or omission) and falls back to TaskPriority.MEDIUM. Tolerating null keeps MCP clients that emit an explicit priority: null from hitting a validation loop. Category is intentionally not a parameter: server-side inference (TaskService._auto_categorize) assigns it from the title and description at creation. To force a different category, use update_task afterwards.

create_task_from_ticket

Create a new task from a RepairShopr ticket.

create_upload_link

Create a one-tap web link for the current user to upload files to Leif. Returns a URL the user opens in any browser (works on mobile) to drop one or more files. Uploaded files are tagged to the calling user and can then be used by other tools (for example, wp_create_post_draft images, or image edits) by referencing their upload_id. The link expires after a short window. After the user uploads, call list_recent_uploads to get the ids.

deadline_* — deadline

deadline_status

Current deadline + stasis status: overdue, due soon (≤7d), due later (≤30d), and stuck deadline-less items. Read-only snapshot for surfacing what’s approaching or has gone quiet.

Parameters not in the seed catalog — see the namespace reference page.

delete_* — delete

delete_file

Delete a local file.

discover_* — discover

discover_services

Discover existing services, clients, and scripts in the codebase.

drive_* — Google Drive

Curated guide: Google Drive.

drive_create_folder

Create a new folder in Google Drive.

drive_create_google_doc

Create a new Google Doc with optional initial content.

drive_get_file_metadata

Get metadata for a Google Drive file.

drive_list_files

List Google Drive files.

drive_move_file

Move a file or folder to a different Google Drive folder.

drive_read_google_doc

Read the text content from a Google Doc.

drive_search_files

Search Google Drive files by filename.

drive_update_google_doc

Write content to an existing Google Doc. mode=‘append’ adds to the end; mode=‘replace’ overwrites all content.

email_* — email

email_intelligence_coverage

Inspect nonsecret stored coverage without ingestion or task creation.

Parameters not in the seed catalog — see the namespace reference page.

email_intelligence_search

Search cited records without Gmail, model calls, or writes.

esa_* — esa

esa_discovery_dry_run

Run the ESA discovery engine in READ-ONLY dry-run mode on nvrbackup. Executes esa_watchlist.discovery_report (the report/dry-run twin of the discovery cron) against the live pricing DB and returns the parsed JSON summary. It writes NOTHING: no esa_discovery_log rows, no Google Sheet append, no watchlist_items changes — safe to run any time. A CSV of every qualified per-(SKU, state) row is written to csv_path on the target host (pull it with transfer_file if you need the fi…

esa_portal_orders_reconcile

Reconcile stored portal order history against local ESA orders. Compares only each successfully synced state’s latest portal snapshot with esa_orders (email-driven intake), joined on (state, normalized line item ID) so identical IDs in different portals cannot collide. Reports: missing_order (active on the portal, never became a local order), unfulfilled_order (active but not fully fulfilled locally), cancelled_but_fulfilled (portal cancelled/refunded while fulfilled locally — refund exposure…

esa_portal_orders_search

Search the stored ESA portal order history and contact info. query matches customer name, email, line item ID, portal order ID, and offer name. Filter with state (e.g. “TX”), status (exact portal status such as “Pending” or “Shipped”), or sku (substring). active_only restricts results to nonterminal lines present in each state’s latest successful snapshot; otherwise both current and historical lines are returned with is_current. Results are newest-synced first with full shipping contact detai…

esa_portal_orders_status

Show ESA portal orders ingest health per state. For each known state portal: effective credential/base-URL status, total historical lines, lines in the current snapshot, active current lines, and latest successful sync metadata. Empty successful snapshots retain their sync timestamp and report zero current/export rows.

Parameters not in the seed catalog — see the namespace reference page.

esa_portal_orders_sync

Sync ESA portal orders.csv exports into the local history table. With no arguments, downloads the full orders CSV live from every state portal with configured credentials (OD_UN / OD_PW) and upserts the complete order history plus customer contact info. Pass state (e.g. “TX”) to sync one portal. To ingest a manually downloaded orders.csv instead, pass state plus either csv_path (a relative path below ESA_PORTAL_ORDERS_IMPORT_ROOT, which defaults to data/imports/esa-portal-orders) or…

events_* — events

events_since

Return episodic events logged since a given timestamp. since is an ISO-8601 timestamp (e.g. 2026-06-10T08:00:00Z). Defaults to the last 24 hours when omitted. Use this to answer “what did Leif do autonomously since my last session?” — covers scheduled job runs, notifications sent, retries, webhook firings, and any other event recorded with log_event. Filter by event_type to narrow results (e.g. job_run, notification, error). Set success_only=true to hide failures.

execute_* — execute

execute_python

Execute Python code in the configured runtime.

extract_* — extract

extract_pdf_tables

Extract tables from a PDF file.

fetch_* — fetch

fetch_rss

Fetch and parse an RSS or Atom feed.

fetch_url

Fetch a URL and return response content or parsed HTML.

finance_* — Finance

Curated guide: Finance.

finance_app_execute_command

Execute a shell command from the remote finance app root directory.

finance_app_file_exists

Check whether a file exists in the remote finance app source tree.

finance_app_list_directory

List files in the remote finance app source tree.

finance_app_read_file

Read a file from the remote finance app source tree.

finance_app_write_file

Write a file in the remote finance app source tree.

finance_get_portfolio

Get the current finance project portfolio summary.

Parameters not in the seed catalog — see the namespace reference page.

finance_get_quote

Get the latest quote for a symbol.

finance_get_risk_limits

Get finance project risk limits.

Parameters not in the seed catalog — see the namespace reference page.

finance_get_risk_status

Get finance project risk status.

Parameters not in the seed catalog — see the namespace reference page.

finance_get_signals

Get recent finance project signals.

finance_health_check

Check the finance project health endpoint.

finance_list_positions

List open positions from the finance project.

finance_list_watchlists

List finance project watchlists.

finance_search_symbols

Search the finance project for symbols.

ga_* — Google Analytics

Curated guide: Google Analytics.

ga_conversion_overview

List top key events/conversions from GA4.

ga_devices

List device-category performance from GA4.

ga_geo_overview

List top countries/regions from GA4.

ga_landing_pages

List top landing pages from GA4.

ga_list_properties

List the configured Google Analytics property aliases available to MCP tools.

Parameters not in the seed catalog — see the namespace reference page.

ga_marketing_overview

Return a marketing-oriented GA4 summary for business lead pages.

ga_realtime_overview

Get a minimal realtime GA4 snapshot.

ga_run_report

Run a generic GA4 report for arbitrary dimensions and metrics.

ga_search_terms

List top site search terms from GA4.

ga_top_pages

List top pages from GA4.

ga_traffic_sources

List top traffic sources from GA4.

gads_* — Google Ads

Curated guide: Google Ads.

gads_account_overview

Return an account-level Google Ads performance summary for the date range.

gads_ad_group_performance

List ad-group-level performance metrics.

gads_campaign_performance

List campaign-level performance metrics (cost, clicks, conversions).

gads_keyword_performance

List keyword-level performance metrics.

gads_list_accessible_customers

List every Google Ads account the configured credentials can access.

Parameters not in the seed catalog — see the namespace reference page.

gads_list_customers

List the configured Google Ads customer aliases available to MCP tools.

Parameters not in the seed catalog — see the namespace reference page.

gads_run_query

Run an arbitrary GAQL (Google Ads Query Language) query for a customer.

gads_search_terms

List the search terms that triggered ads, with performance metrics.

gads_update_ad_group_bid

Set an ad group’s maximum CPC bid (cpc_bid in dollars, e.g. 2.50). Set validate_only=true to dry-run the change without applying it.

gads_update_ad_group_status

Pause, enable, or remove an ad group (status: ENABLED, PAUSED, REMOVED). Set validate_only=true to dry-run the change without applying it.

gads_update_ad_status

Pause, enable, or remove a single ad (status: ENABLED, PAUSED, REMOVED). Set validate_only=true to dry-run the change without applying it.

gads_update_campaign_status

Pause, enable, or remove a campaign (status: ENABLED, PAUSED, REMOVED). Set validate_only=true to dry-run the change without applying it.

gbp_* — Google Business Profile

Curated guide: Google Business Profile.

gbp_get_location

Fetch a single Google Business Profile location.

gbp_list_accounts

List accessible Google Business Profile accounts.

gbp_list_locations

List locations for a Google Business Profile account.

gbp_local_overview

Aggregate local performance across all accessible GBP locations.

gbp_location_performance

Fetch local performance metrics for one Google Business Profile location.

gbp_multi_location_summary

Aggregate GBP performance across multiple locations.

get_* — get

get_bill_entry_batch

Return durable bill-entry status without external reads or writes.

get_check_deposit_batch

Return durable check-deposit status without external reads or writes.

get_daily_review

Return a prioritized daily task review.

Parameters not in the seed catalog — see the namespace reference page.

get_excel_info

Get workbook sheet metadata.

get_location

Get the user’s current location information.

Parameters not in the seed catalog — see the namespace reference page.

get_note

Fetch a single note’s full content by its ID.

get_pdf_info

Get metadata and information about a PDF file.

get_person_summary

Summarize the relationship history with a specific person.

get_project

Return a single project by ID.

get_project_context

Read a project’s combined context file contents.

get_project_status

Return context quality information for a project.

get_python_namespace

Show variables currently defined in the Python execution namespace.

Parameters not in the seed catalog — see the namespace reference page.

get_recent_notes

Return the most recent notes.

get_time

Get the current time in Wayne’s timezone (US/Central) with day-of-week and time-of-day bucket. Returns a JSON object with: - iso: ISO 8601 timestamp in CT - human: human-friendly formatted time - day_of_week: full day name (e.g., ‘Tuesday’) - hour_24: hour of day (0-23) in CT - bucket: time-of-day bucket per leif_self.md — one of: late_night, morning, midday, maintenance_window, evening - is_maintenance_window: True if 14

Parameters not in the seed catalog — see the namespace reference page.

get_topic_overview

Summarize discussion history for a topic.

get_weather

Get a weather forecast for a location.

github_* — GitHub

Curated guide: GitHub.

github_batch_move_files

Move multiple files in GitHub.

github_create_issue

Create a GitHub issue.

github_create_or_update_file

Create or update a file in GitHub.

github_get_file

Fetch a file from a GitHub repository.