Macro is a unified workspace for teams: email, chat, docs, tasks, agents, calls, and CRM — @-linked together with shared AI memory.
TypeScript agentaiai-agentsall-in-onecrmcrm-systememaillinearmcpmessagingnotesnotionofficerustslackslack-alternativestartupstartup-toolstypescriptworkspace
fix(import): slack import change (#5152)
teo73c5f9e0464d
diff --git a/crates/import/src/domain/service/prompts.rs b/crates/import/src/domain/service/prompts.rsindex e4d7fcec7e..5ac34588fa 100644--- a/crates/import/src/domain/service/prompts.rs+++ b/crates/import/src/domain/service/prompts.rs@@ -10,25 +10,39 @@ use crate::domain::models::{ImportEntity, ImportSource, metadata_label};/// System prompt for a gather session over one connector.pub fn gather_system(source: ImportSource) -> String {- let (what, foreign_id_rule, metadata_hint) = match source {+ let (what, foreign_id_rule, metadata_hint, search_strategy) = match source {ImportSource::Linear => ("the user's most relevant recent Linear issues","the Linear issue identifier (e.g. `ENG-142`); fall back to the issue URL only when \no identifier exists","identifier, title, description (short markdown), status, priority, assignee, \assignee_email, due_date (ISO date, when set), url",+ "Always pass a non-empty query to search tools.",),ImportSource::Notion => ("the user's most recently updated, substantive Notion pages","the page URL (it is normalized to the stable page id server-side)","title, url, summary (one line). Do NOT fetch or include page content — content is \fetched later, only for pages the user accepts",+ "Always pass a non-empty query to search tools.",),ImportSource::Slack => ("the Slack channels the user is most active in","the Slack channel id (e.g. `C0123456789`); fall back to the channel name","name (without the leading #), channel_id, purpose, participants (name + email when \available)",+ "Slack-specific discovery strategy:\n\+ - FIRST call `Search channels` with an explicitly empty search string: \+ `{\"query\": \"\"}`. For this Slack MCP tool, an empty query lists all channels the \+ connected user can see. Do not substitute `active`, `recent`, `all`, or `*`.\n\+ - Follow the result's pagination cursor when one is present until you have enough \+ candidates or there are no more pages.\n\+ - Prefer channels whose names, topics, or purposes indicate substantive work. You \+ may use one broad `Search messages & files` call afterward to rank the enumerated \+ channels by recent activity, but do not use message search as a prerequisite for \+ discovering that a channel exists.\n\+ - Participant details are optional. Do not drop a channel merely because member \+ names or emails are unavailable.",),};format!(@@ -37,8 +51,9 @@ pub fn gather_system(source: ImportSource) -> String {\n\Use the connected tools to find 8-15 strong candidates: recently active, substantive, \and clearly relevant to the user's own work. Prefer one or two broad searches/list \- calls over many narrow ones. Always pass a non-empty query and only the parameters a \- tool's schema requires; if a call fails validation, fix the arguments and retry once.\n\+ calls over many narrow ones. Pass only parameters the tool's schema supports; if a call \+ fails validation, fix the arguments and retry once.\n\+ {search_strategy}\n\\n\For EACH candidate, call `CreateImportEntity` once with:\n\- `foreign_id`: {foreign_id_rule}\n\diff --git a/crates/import/src/domain/service/test.rs b/crates/import/src/domain/service/test.rsindex 7f497dbab7..24a7d3dd29 100644--- a/crates/import/src/domain/service/test.rs+++ b/crates/import/src/domain/service/test.rs@@ -100,6 +100,17 @@ fn notion_import_turns_scale_with_pages_and_cap() {assert_eq!(notion_import_max_turns(100), 40);}+#[test]+fn slack_gather_lists_channels_with_an_explicit_empty_query() {+ let prompt = prompts::gather_system(ImportSource::Slack);++ assert!(prompt.contains("FIRST call `Search channels`"));+ assert!(prompt.contains(r#"{"query": ""}"#));+ assert!(prompt.contains("an empty query lists all channels"));+ assert!(!prompt.contains("Always pass a non-empty query"));+ assert!(prompt.contains("Participant details are optional"));+}+#[test]fn notion_fetch_text_parses_the_fetch_document_shape() {let (title, body) = parse_notion_fetch_text(