> ## Documentation Index
> Fetch the complete documentation index at: https://developers.everflow.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

> Latest updates and changes to the Everflow API and documentation.

This page tracks notable changes to the Everflow API, SDKs, and developer documentation. Check back regularly for updates on new endpoints, breaking changes, and documentation improvements.

<Update label="2026-07-07" tags={["MCP"]}>
  ## MCP Server — reporting trust and read-tool refinements

  * **Conversions by transaction** — `list_entities(type="conversion", filters={"transaction_id":…})` returns all conversions for a transaction as a flat, paginated list (no date range). See [Events & Attribution](/ai-automation/mcp/tools/events#conversion).
  * **Creative markup** — `get_entity(type="creative")` returns the full record including `html_code`; `list_entities` stays lean. See [Supported entity types](/ai-automation/mcp/tools/generic#supported-entity-types).
  * **`advertiser_event` by offer** — `list_entities(type="advertiser_event")` accepts `offer_id` in place of `advertiser_id` (the advertiser is resolved from the offer).
  * **`applied_query`** — `run_performance_report` and `run_network_summary` echo the resolved query (date window, timezone, currency, filters) so results are reproducible. See [run\_performance\_report](/ai-automation/mcp/tools/reporting#run_performance_report).
  * **`metric_glossary`** — both reporting tools include inline definitions for the easily-misread metrics (the conversion-counting family, event revenue, percent vs ratio).
  * **Server-side `totals`** — `run_performance_report` returns a `totals` object over the full matched set, with ratios re-derived from the sums rather than averaged — so you never sum rows yourself.
  * **Capability onboarding** — the server now advertises what it can do (and starter questions) to the client on connect, so "what can you help me with?" gets a grounded answer. See [Quickstart](/ai-automation/mcp/quickstart#first-questions-to-try).
</Update>

<Update label="2026-06-29" tags={["API", "MCP"]}>
  ## Traffic Health — Network API endpoints and MCP tools

  Traffic Health — the operational health of the domains and hosting IPs behind your tracking and conversion links (uptime, SSL, DNS and expiry incidents, blocklist reputation, remediation tasks, and per-domain hosting/SSL, assignment and revenue detail) — is now available through both the **Network API** and the **MCP Server**.

  * **Network API** — a full Traffic Health reference covering network-wide and per-domain situation, incidents, tasks, blocklist reputation, assignments, usage, and mismatches. Start with [Get Traffic Health Access](/api-reference/get-networkstrafficaccess).
  * **MCP Server** — two read-only tools: **`list_traffic_health`** for network-wide rollups and **`get_traffic_health_domain`** for one domain's full picture in a single call. See [Traffic Health tools](/ai-automation/mcp/tools/traffic-health).
  * **Free vs Premium** — core health, tasks, hosting and SSL are available to any network with Traffic Health enabled; blocklist reputation, external domain monitoring, and per-domain assignments/usage require **Traffic Health Premium**. Premium-only requests return a notice instead of data rather than erroring.
</Update>

<Update label="2026-06-15" tags={["MCP"]}>
  ## MCP Server — continuous improvements

  A batch of refinements to the read-only MCP tools, focused on counting, scoped summaries, metric clarity, and timezone correctness.

  * **`count_entities`** — a new generic tool that returns **only the match count** for a type + filter set, with no records or pagination. Use it for "how many" questions instead of paging `list_offers` / `list_affiliates` / `list_entities` and tallying. See [count\_entities](/ai-automation/mcp/tools/generic#count_entities).
  * **`run_network_summary` entity filters** — scope the headline totals to specific `offer`, `affiliate`, `advertiser`, `creative`, or `campaign` (pipe for OR). Combined with `include=comparison`, this answers "is offer 91 up or down vs last week?" — the filter is applied to the comparison period too. See [run\_network\_summary](/ai-automation/mcp/tools/reporting#run_network_summary).
  * **`get_report_schema` metric definitions** — the response now includes `metric_definitions`, one entry per metric with `{ name, description, unit, formula, is_custom }`, including this network's **user-defined custom metrics**. It is the source of truth for "what metrics exist" and "how is `<metric>` calculated". See [get\_report\_schema](/ai-automation/mcp/tools/reporting#get_report_schema).
  * **Metric filters & percent metrics** — `run_performance_report` accepts `metric_filters` (e.g. `clicks>100,cvr<2`, `invalid_clicks<=0`) on aggregated rows. `cvr`, `ctr`, and `margin` are percentages (`2.84` = 2.84%); `roas` is a ratio. Metric descriptions/formulas are now exact.
  * **Network timezone & currency defaults** — `run_performance_report`, `run_network_summary`, and `search_activity` now default to the **network's timezone and base currency** (matching the Everflow portal) instead of UTC/USD when those parameters are omitted.
  * **Raw-event truncation hints** — when `search_activity` hits its cap it returns `truncated` (and `total_matching` for clicks) plus a `note` steering to `run_performance_report` (e.g. `dimensions=conversion_error_code`) for counts and breakdowns.
</Update>

<Update label="2026-06-10" tags={["API", "MCP"]}>
  ## Sub and advertiser parameters extended from 5 to 10

  Tracking now supports **`sub1`–`sub10`** affiliate sub-parameters and **`adv1`–`adv10`** advertiser parameters everywhere the first five were accepted. Existing integrations are unaffected — the new parameters are optional and omitted from payloads when empty.

  The expansion applies across every surface:

  * **Tracking links and API endpoints** — click and conversion endpoints accept and return the full range (600-char limit per value, unchanged)
  * **Reporting** — available as filters and per-record fields on clicks, conversions, impressions, exports, and scheduled reports
  * **JavaScript SDK** — [`EF.click()`](/sdk/click-tracking), [`EF.conversion()`](/sdk/conversion-tracking), and [`EF.impression()`](/sdk/impression-tracking) accept the new parameters
  * **Webhooks & Firehose** — [conversion webhook](/webhooks/conversion-webhooks) and [Firehose](/user-guide/firehose) payloads carry `sub1`–`sub10` and `adv1`–`adv10`
  * **MCP Server** — `run_performance_report`, `search_activity`, and the event entity types filter on the full range; see [Events & Attribution](/ai-automation/mcp/tools/events), the new home for click, conversion, and transaction lookup documentation
</Update>

<Update label="2026-06-08" tags={["MCP"]}>
  ## MCP Server — transaction lookup and tool consolidation

  One capability added and four tools consolidated, with no net growth in the tool list.

  * **Transaction lookup** — `get_entity(type="transaction", id="<transaction ID>")` returns the full attribution chain in one call: the click, its conversions (each with `attribution_method`), pixel-fire logs, and on-hold conversions. See [get\_entity](/ai-automation/mcp/tools/generic#get_entity).
  * **`search_activity`** replaces `search_clicks` and `search_conversions` — set `type` to `click` or `conversion`. Same filters and data.
  * **Deprecations** — `get_click`, `get_conversion`, `search_clicks`, and `search_conversions` are now served via `get_entity` / `search_activity`. They remain **fully callable** for backward compatibility but are no longer listed. See [Deprecated tools](/ai-automation/mcp/tools#deprecated-tools) for the migration table.
</Update>

<Update label="2026-06-02" tags={["MCP"]}>
  ## MCP Server — multi-value filters for performance reports

  `run_performance_report` now matches **multiple IDs for a single filter in one query**. Separate values with a pipe — `offer:1|2|5` — or repeat the key (`offer:1,offer:2`); both return all of the listed entities together instead of requiring a separate report per ID.

  * Supported on the ID filters: `offer`, `offer_group`, `affiliate`, `advertiser`, `creative`, `campaign`, `category`, `network`, `tracking_domain`, `channel`
  * See the [run\_performance\_report filter reference](/ai-automation/mcp/tools/reporting#run_performance_report) and the [Compare a specific set of offers](/ai-automation/mcp/examples#compare-a-specific-set-of-offers) example
</Update>

<Update label="2026-05-19" tags={["MCP"]}>
  ## MCP Server — new entity types

  7 additional types are now supported by `get_entity`, `list_entities`, and `get_entity_schema`: `employee`, `category`, `channel`, `label`, `affiliate_tier`, `offer_url`, and `advertiser_event`. The total number of supported entity types is now 27.

  * View the updated [entity type reference](/ai-automation/mcp/tools/generic#get_entity_schema) for required filters and usage notes per type
</Update>

<Update label="2026-04-30" tags={["API"]}>
  ## Update Conversion Notes endpoint documented

  Documentation added for the existing `PUT /networks/reporting/conversions/{conversionId}/notes` endpoint, which lets you attach free-form notes to a conversion. The endpoint itself is unchanged — only the docs are new.

  * View the [Update Conversion Notes](/api-reference/put-networksreportingconversionsnotes) endpoint reference (Network API → Conversion Updates)
  * Pass an empty string to clear existing notes
  * Returns the **full updated conversion record** in the response
</Update>

<Update label="2026-04-23" tags={["MCP"]}>
  ## MCP Server documentation added

  A new **MCP** tab has been added to the developer hub covering the Everflow Model Context Protocol (MCP) Server — a native integration for AI assistants (Claude, Cursor, Windsurf, and others) that gives LLMs direct, structured access to your Everflow data without custom API wrappers.

  * **Overview** — what the MCP Server is and how it works
  * **Quickstart** — step-by-step setup for Claude Desktop, Cursor, and Windsurf
  * **Tools reference** — all 16 read-only tools with full parameter, dimension, filter, and metric tables
  * **Prompt templates** — ready-to-use prompts for common workflows
  * **Examples** — real conversational walkthroughs
  * **FAQ** — common questions and troubleshooting
</Update>