Skip to content

YC (S26) | Open Computer History | Continuously record your company computer work, map your workflows, help you find work worth automating, and power your agents' context

simplify: let the summary agent choose and save meeting titles

Louis Beaumontaddf80875d05
diff --git a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-summary-stream.test.ts b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-summary-stream.test.ts
index 37df76a46..43332a209 100644
--- a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-summary-stream.test.ts
+++ b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-summary-stream.test.ts
@@ -26,12 +26,6 @@ const textStart = (contentIndex = 0) => ({
});
describe("meeting summary stream", () => {
- it("preserves the topic heading through the streamed-to-saved handoff", () => {
- const markdown = "### Launch planning and release ownership\nWe agreed to ship Tuesday.";
- let state = advanceMeetingSummaryStream(emptyMeetingSummaryStream(), textEnd(`## Summary\n${markdown}`));
- state = advanceMeetingSummaryStream(state, { type: "tool_execution_start" });
- expect(state.markdown).toBe(markdown);
- });
it("streams only markdown after the summary heading", () => {
let state = emptyMeetingSummaryStream();
state = advanceMeetingSummaryStream(
diff --git a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.test.tsx b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.test.tsx
index 88a6e005b..5c64126fe 100644
--- a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.test.tsx
+++ b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.test.tsx
@@ -17,14 +17,6 @@ import {
stopMeetingAndOpenSummary,
} from "./meeting-workspace";
-it("keeps the recoverable topic heading but requires a real summary body", () => {
- const topic = "### Launch planning and release ownership";
- expect(extractMeetingSummary(`## Summary\n${topic}\nWe agreed to ship Tuesday.`))
- .toBe(`${topic}\nWe agreed to ship Tuesday.`);
- expect(extractMeetingSummary(`## Summary\n${topic}`)).toBeNull();
- expect(extractMeetingSummary(`## Summary\n${topic}\nI couldn't summarize this meeting.`)).toBeNull();
-});
-
it("does not treat a saved refusal as a usable summary", () => {
const note = "## Summary\nI couldn't produce a reliable summary for this meeting. The recorded evidence appears mismatched.";
expect(extractMeetingSummary(note)).toBeNull();
diff --git a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.tsx b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.tsx
index 309286068..06ab10aa4 100644
--- a/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.tsx
+++ b/apps/screenpipe-app-tauri/components/meeting-notes/meeting-workspace.tsx
@@ -169,16 +169,13 @@ export function extractMeetingSummary(markdown: string): string | null {
const body = markdown.slice(latest.index + latest[0].length).trim();
// A refusal is an error outcome, even when an older run saved it under
// the Summary heading. Match the server's narrow failure-prefix check.
- const content = body.startsWith("### ")
- ? body.slice(body.indexOf("\n") < 0 ? body.length : body.indexOf("\n") + 1).trim()
- : body;
- const normalized = content.toLowerCase().replace(/’/g, "'");
+ const normalized = body.toLowerCase().replace(/’/g, "'");
const failurePrefixes = [
"i couldn't produce a reliable summary", "i could not produce a reliable summary",
"i couldn't summarize", "i could not summarize", "i can't summarize", "i cannot summarize",
"unable to summarize this meeting", "no speech was captured", "no transcript was available",
];
- return content && !failurePrefixes.some((prefix) => normalized.startsWith(prefix)) ? body : null;
+ return body && !failurePrefixes.some((prefix) => normalized.startsWith(prefix)) ? body : null;
}
/**
diff --git a/crates/screenpipe-core/assets/pipes/meeting-summary/pipe.md b/crates/screenpipe-core/assets/pipes/meeting-summary/pipe.md
index bf101f974..6a34b35f9 100644
--- a/crates/screenpipe-core/assets/pipes/meeting-summary/pipe.md
+++ b/crates/screenpipe-core/assets/pipes/meeting-summary/pipe.md
@@ -120,7 +120,7 @@ every speaker used in the summary must have either an evidence-backed participan
persist a real name only when the meeting-local mapping resolves a specific nonzero speaker id. never rename speaker id `0`, and never overwrite one persistent id that appears on conflicting input/output sides; keep those corrections meeting-local in the summary. note which persistent speakers you renamed and which names are meeting-local only in your final message.
-step 3 — write the summary out as your own message, before you save it. this message must contain no tool call; end the turn after it. start a line with exactly `## Summary` and put the finished summary markdown after that heading. The first nonempty line of that markdown must be `### <TOPIC_TITLE>`, with a specific 5-8 word title grounded in this meeting's transcript (at most 120 characters, no quotes or "meeting about" prefix). Then write the summary body. Always include that topic heading, even when the record already has a descriptive name: it lets the server recover the generated title if the save step fails. Keep the heading in `<YOUR_SUMMARY>` when saving. the meeting UI streams this section live while you write it — it is the only way the user sees anything before the run ends — and it is the same markdown you pass as `<YOUR_SUMMARY>` in step 3b.
+step 3 — write the summary out as your own message, before you save it. this message must contain no tool call; end the turn after it. start a line with exactly `## Summary` and put the finished summary markdown after that heading. the meeting UI streams this section live while you write it — it is the only way the user sees anything before the run ends — and it is the same markdown you pass as `<YOUR_SUMMARY>` in step 3b.
this step is not optional and it is not the closing report. "meeting 112 was summarized and saved to its record" is a report; it does not satisfy this step. saving a summary you never printed means the user watched a placeholder for the entire run and then got nothing to read, so treat that as a failed run. keep planning, tool narration, and save confirmations out of the section after the heading — those belong in your closing message in step 4.
@@ -138,7 +138,7 @@ step 3b — now save it through the dedicated summary endpoint. the server merge
-H "Content-Type: application/json" \
--data @/tmp/summary.json
-`-f` matters: if this call fails, say so in your closing message instead of reporting success. for the title: pass the same topic title used in the first `###` heading. The server renames only missing or placeholder titles ("untitled", "meeting", or just the app name), and preserves descriptive user or calendar titles, including edits made while you were summarizing. if there's nothing useful to summarize (empty transcript, irrelevant audio), say so out loud and skip the save — don't write a placeholder.
+`-f` matters: if this call fails, say so in your closing message instead of reporting success. Naming the meeting is part of summarizing it: choose a concise, specific title from what was actually discussed, rather than leaving it named after the calling app. Before preparing `/tmp/title.txt`, fetch `GET /meetings/<MEETING_ID>` again and compare its title with the record from step 1. If the title changed while you were working, or already describes the discussion, leave `/tmp/title.txt` empty to preserve it. Otherwise put your chosen title in that file and save it together with the summary through the existing endpoint above. Do not add a special title heading to the summary or make a separate AI request for naming. After saving, read `GET /meetings/<MEETING_ID>` and verify that the summary and any title you supplied are present. If either is missing, report the failed save instead of claiming success. If there is nothing useful to summarize, skip the save and leave the title unchanged.
step 4 — offer to push the summary into one of the user's connected apps (ask, never push on your own). list what's actually connected, then let them pick with one click:
diff --git a/crates/screenpipe-core/src/pipes/builtin_migrations.rs b/crates/screenpipe-core/src/pipes/builtin_migrations.rs
index 98b4b4254..9b2f02b12 100644
--- a/crates/screenpipe-core/src/pipes/builtin_migrations.rs
+++ b/crates/screenpipe-core/src/pipes/builtin_migrations.rs
@@ -285,9 +285,7 @@ replace `<EXISTING_NOTE>` with the meeting's current `note` field (empty string
const LEGACY_MEETING_AUDIO_FETCH: &str = r###" curl -s -G -H "$A" --data-urlencode "start_time=$S" --data-urlencode "end_time=$E" \
-d content_type=audio -d limit=500 "http://localhost:3030/search" -o /tmp/audio.json &"###;
-const LEGACY_SUMMARY_OUTPUT_WITHOUT_TITLE: &str = r###"step 3 — write the summary out as your own message, before you save it. this message must contain no tool call; end the turn after it. start a line with exactly `## Summary` and put the finished summary markdown after that heading. the meeting UI streams this section live while you write it — it is the only way the user sees anything before the run ends — and it is the same markdown you pass as `<YOUR_SUMMARY>` in step 3b."###;
-
-const LEGACY_SUMMARY_TITLE_RULE: &str = r###"`-f` matters: if this call fails, say so in your closing message instead of reporting success. for the title: if the current title is missing, generic ("untitled", "meeting", just the app name) or doesn't capture what actually happened, pass a 5-8 word plain-english title (no quotes, no "meeting about…" prefix) — otherwise pass the empty string so a user-set title is left alone. if there's nothing useful to summarize (empty transcript, irrelevant audio), say so out loud and skip the save — don't write a placeholder."###;
+const LEGACY_MEETING_TITLE_SAVE_RULE: &str = r###"`-f` matters: if this call fails, say so in your closing message instead of reporting success. for the title: if the current title is missing, generic ("untitled", "meeting", just the app name) or doesn't capture what actually happened, pass a 5-8 word plain-english title (no quotes, no "meeting about…" prefix) — otherwise pass the empty string so a user-set title is left alone. if there's nothing useful to summarize (empty transcript, irrelevant audio), say so out loud and skip the save — don't write a placeholder."###;
/// Swaps for `meeting-summary`, oldest defect first.
fn meeting_summary_swaps() -> Vec<FragmentSwap> {
@@ -556,23 +554,16 @@ fn meeting_summary_swaps() -> Vec<FragmentSwap> {
});
}
- for (old, start, end) in [
- (
- LEGACY_SUMMARY_OUTPUT_WITHOUT_TITLE,
- "step 3 — write the summary",
- "\n\nthis step is not optional",
- ),
- (LEGACY_SUMMARY_TITLE_RULE, "`-f` matters:", "\n\nstep 4"),
- ] {
- if let Some(new) =
- section_between(bundled_prompt("meeting-summary").unwrap_or(""), start, end)
- {
- swaps.push(FragmentSwap {
- why: "keep the AI topic in the recoverable summary and preserve descriptive meeting names",
- old,
- new,
- });
- }
+ if let Some(rule) = section_between(
+ bundled_prompt("meeting-summary").unwrap_or(""),
+ "`-f` matters:",
+ "\n\nstep 4",
+ ) {
+ swaps.push(FragmentSwap {
+ why: "have the summary agent choose a useful title and verify its save",
+ old: LEGACY_MEETING_TITLE_SAVE_RULE,
+ new: rule,
+ });
}
swaps
@@ -956,29 +947,25 @@ fn replace_prompt_body_when_hash_matches(
#[cfg(test)]
mod tests {
#[test]
- fn installs_recoverable_topic_titles_without_changing_user_configuration() {
+ fn migrates_meeting_title_instructions_without_changing_user_settings() {
let prompt = bundled("meeting-summary");
- let output = section_between(
- prompt,
- "step 3 — write the summary",
- "\n\nthis step is not optional",
- )
- .unwrap();
let rule = section_between(prompt, "`-f` matters:", "\n\nstep 4").unwrap();
let stale = format!(
"{}\nMy custom instructions stay here.\n",
prompt
- .replace(output, LEGACY_SUMMARY_OUTPUT_WITHOUT_TITLE)
- .replace(rule, LEGACY_SUMMARY_TITLE_RULE)
+ .replace(rule, LEGACY_MEETING_TITLE_SAVE_RULE)
.replace("enabled: true", "enabled: false")
);
let migrated = migrate_builtin_pipe_text("meeting-summary", &stale).unwrap();
- assert!(migrated.contains("### <TOPIC_TITLE>"));
- assert!(migrated.contains("preserves descriptive user or calendar titles"));
+ assert!(migrated.contains("Naming the meeting is part of summarizing it"));
+ assert!(migrated.contains("If the title changed while you were working"));
+ assert!(migrated.contains("verify that the summary and any title you supplied are present"));
+ assert!(migrated.contains("skip the save and leave the title unchanged"));
assert!(migrated.contains("enabled: false"));
assert!(migrated.ends_with("My custom instructions stay here.\n"));
assert!(migrate_builtin_pipe_text("meeting-summary", &migrated).is_none());
assert!(migrate_builtin_pipe_text("day-recap", &stale).is_none());
+ assert!(migrate_builtin_pipe_text("meeting-summary", "My custom summary prompt").is_none());
}
#[test]
diff --git a/crates/screenpipe-db/src/db/meetings.rs b/crates/screenpipe-db/src/db/meetings.rs
index cf41724ac..7d9403b73 100644
--- a/crates/screenpipe-db/src/db/meetings.rs
+++ b/crates/screenpipe-db/src/db/meetings.rs
@@ -1460,31 +1460,6 @@ impl DatabaseManager {
Ok(())
}
- /// Save a summary while preserving a title edited since the caller read it.
- /// The comparison and write share the existing coordinated transaction.
- pub async fn update_meeting_summary(
- &self,
- id: i64,
- expected_title: Option<&str>,
- generated_title: Option<&str>,
- note: &str,
- ) -> Result<(), SqlxError> {
- let mut tx = self.begin_immediate_with_retry().await?;
- sqlx::query(
- "UPDATE meetings SET note = ?1,
- title = CASE WHEN title IS ?2 AND ?3 IS NOT NULL THEN ?3 ELSE title END
- WHERE id = ?4",
- )
- .bind(note)
- .bind(expected_title)
- .bind(generated_title)
- .bind(id)
- .execute(&mut **tx.conn())
- .await?;
- tx.commit().await?;
- Ok(())
- }
-
/// Merge `ids` into the lowest-id survivor.
///
/// Preserves user-entered metadata across the merge:
diff --git a/crates/screenpipe-engine/src/meeting_summary/finalizer.rs b/crates/screenpipe-engine/src/meeting_summary/finalizer.rs
index 4952569e6..f828b30d1 100644
--- a/crates/screenpipe-engine/src/meeting_summary/finalizer.rs
+++ b/crates/screenpipe-engine/src/meeting_summary/finalizer.rs
@@ -497,14 +497,11 @@ mod db_tests {
let (_dir, db) = test_db().await;
let store = SqlitePipeStore::new(db.clone());
let meeting_id = ended_meeting(&db).await;
- db.update_meeting(meeting_id, None, None, Some("Zoom"), None, None, None)
- .await
- .unwrap();
let run = completed_run(
&store,
&meeting_id.to_string(),
&agent_end_stdout(
- "## Summary\n### Rollout planning and follow-up ownership\nWe agreed on the rollout plan and booked a follow-up for Thursday.",
+ "## Summary\nWe agreed on the rollout plan and booked a follow-up for Thursday.",
),
)
.await;
@@ -514,11 +511,7 @@ mod db_tests {
let meeting = db.get_meeting_by_id(meeting_id).await.unwrap();
let note = meeting.note.unwrap();
- assert!(note.starts_with("## Summary\n### Rollout planning and follow-up ownership"));
- assert_eq!(
- meeting.title.as_deref(),
- Some("Rollout planning and follow-up ownership")
- );
+ assert!(note.starts_with("## Summary\nWe agreed on the rollout plan"));
// The run stays a success — the outcome was repaired, not the run.
assert_eq!(
execution_state(&db, run).await,
diff --git a/crates/screenpipe-engine/src/meeting_summary/notes.rs b/crates/screenpipe-engine/src/meeting_summary/notes.rs
index 9ca325fdc..7f9d2a291 100644
--- a/crates/screenpipe-engine/src/meeting_summary/notes.rs
+++ b/crates/screenpipe-engine/src/meeting_summary/notes.rs
@@ -26,13 +26,7 @@ pub(crate) fn is_summary_heading(line: &str) -> bool {
/// Clear failure messages are outcomes to recover from, not meeting summaries.
/// Keep this narrow: a genuine summary may mention missing evidence or errors.
pub(crate) fn summary_has_content(summary: &str) -> bool {
- let summary = summary.trim();
- let body = if summary.starts_with("### ") {
- summary.split_once('\n').map(|(_, body)| body).unwrap_or("")
- } else {
- summary
- };
- let normalized = body.trim().to_lowercase().replace('’', "'");
+ let normalized = summary.trim().to_lowercase().replace('’', "'");
!normalized.is_empty()
&& ![
"i couldn't produce a reliable summary",
@@ -87,51 +81,6 @@ pub(crate) fn merge_summary_into_note(existing_note: Option<&str>, summary: &str
}
}
-/// Only placeholders are eligible for automatic naming. A descriptive title may
-/// have come from the user or calendar and must survive subsequent summaries.
-fn is_placeholder_title(title: Option<&str>, meeting_app: &str) -> bool {
- let title = title.unwrap_or("").trim().to_lowercase();
- title.is_empty()
- || title == meeting_app.trim().to_lowercase()
- || matches!(
- title.as_str(),
- "untitled"
- | "untitled meeting"
- | "meeting"
- | "google meet"
- | "zoom"
- | "microsoft teams"
- | "teams"
- | "slack"
- | "facetime"
- | "webex"
- )
-}
-
-/// The summary contract puts the generated topic in its first H3. Keeping it
-/// in the printed summary lets the finalizer recover both fields after a failed
-/// agent save. Older summaries without this heading remain valid.
-fn summary_title(summary: &str) -> Option<&str> {
- summary
- .trim_start()
- .lines()
- .next()?
- .strip_prefix("### ")
- .map(str::trim)
- .filter(|title| !title.is_empty() && title.chars().count() <= 120)
- .filter(|title| {
- !matches!(
- title.to_lowercase().as_str(),
- "summary"
- | "overview"
- | "key topics"
- | "key decisions"
- | "action items"
- | "next steps"
- )
- })
-}
-
/// Persist a finished summary (and optional title) onto a meeting record.
///
/// One server-side write path shared by the HTTP endpoint and the run
@@ -158,16 +107,8 @@ pub(crate) async fn save_meeting_summary(
let title = title
.map(str::trim)
.filter(|t| !t.is_empty())
- .filter(|t| {
- t.chars().count() <= 120
- && !t.contains(['\r', '\n'])
- && !is_placeholder_title(Some(t), &meeting.meeting_app)
- })
- .or_else(|| {
- summary_title(summary).filter(|t| !is_placeholder_title(Some(t), &meeting.meeting_app))
- })
- .filter(|_| is_placeholder_title(meeting.title.as_deref(), &meeting.meeting_app));
- db.update_meeting_summary(id, meeting.title.as_deref(), title, &note)
+ .map(str::to_string);
+ db.update_meeting(id, None, None, title.as_deref(), None, Some(&note), None)
.await
.map_err(|e| e.to_string())?;
db.get_meeting_by_id(id)
@@ -249,19 +190,6 @@ mod tests {
.unwrap();
assert!(save_meeting_summary(&db, id, " \n", None).await.is_err());
- assert!(
- save_meeting_summary(&db, id, "### A title with no summary", None)
- .await
- .is_err()
- );
- assert!(save_meeting_summary(
- &db,
- id,
- "### A generated topic title\nI couldn't summarize this meeting.",
- None
- )
- .await
- .is_err());
let saved = save_meeting_summary(&db, id, "Decisions were made.", Some("Pricing sync"))
.await
@@ -285,103 +213,4 @@ mod tests {
assert_eq!(refreshed.note.as_deref(), Some("## Summary\nRefreshed."));
assert_eq!(refreshed.title.as_deref(), Some("Pricing sync"));
}
- #[test]
- fn title_contract_preserves_descriptive_names_and_ignores_old_section_headings() {
- for title in [
- None,
- Some(""),
- Some(" Untitled meeting "),
- Some("GOOGLE MEET"),
- Some("Zoom"),
- ] {
- assert!(is_placeholder_title(title, "google meet"));
- }
- assert!(!is_placeholder_title(
- Some("Launch review with design"),
- "google meet"
- ));
- assert_eq!(
- summary_title("### Launch planning and release ownership\nWe agreed on owners."),
- Some("Launch planning and release ownership")
- );
- for summary in [
- "A normal older summary.",
- "### Action items\n- Ship",
- "### ",
- "## Summary",
- ] {
- assert_eq!(summary_title(summary), None);
- }
- }
-
- #[tokio::test]
- async fn recovers_topic_title_and_preserves_names_on_refresh() {
- let dir = tempfile::tempdir().unwrap();
- let db = DatabaseManager::new(
- &dir.path().join("titles.db").to_string_lossy(),
- Default::default(),
- )
- .await
- .unwrap();
- let summary = "### Launch planning and release ownership\nThe team agreed to ship on Tuesday and assigned review owners.";
- for original in [
- None,
- Some("google meet"),
- Some("Untitled meeting"),
- Some("Launch review with design"),
- ] {
- let id = db
- .insert_meeting("google meet", "audio_process", None, None)
- .await
- .unwrap();
- db.end_meeting(id, &chrono::Utc::now().to_rfc3339(), None)
- .await
- .unwrap();
- db.update_meeting(
- id,
- None,
- None,
- original,
- None,
- Some("My preparation notes"),
- None,
- )
- .await
- .unwrap();
- let saved = save_meeting_summary(&db, id, summary, None).await.unwrap();
- let expected = if original == Some("Launch review with design") {
- original
- } else {
- Some("Launch planning and release ownership")
- };
- assert_eq!(saved.title.as_deref(), expected);
- assert!(saved
- .note
- .unwrap()
- .starts_with("My preparation notes\n\n## Summary"));
- let refreshed = save_meeting_summary(
- &db,
- id,
- "Updated summary.",
- Some("A different generated title"),
- )
- .await
- .unwrap();
- assert_eq!(refreshed.title.as_deref(), expected);
- }
- let id = db
- .insert_meeting("Zoom", "audio_process", None, None)
- .await
- .unwrap();
- db.update_meeting(id, None, None, Some("My new name"), None, None, None)
- .await
- .unwrap();
- // Simulate a rename between the summary's read and write.
- db.update_meeting_summary(id, None, Some("AI topic"), summary)
- .await
- .unwrap();
- let saved = db.get_meeting_by_id(id).await.unwrap();
- assert_eq!(saved.title.as_deref(), Some("My new name"));
- assert_eq!(saved.note.as_deref(), Some(summary));
- }
}
diff --git a/docs/COVERAGE.md b/docs/COVERAGE.md
index 1017305a1..8bfd16266 100644
--- a/docs/COVERAGE.md
+++ b/docs/COVERAGE.md
@@ -38,15 +38,25 @@ results and `cargo llvm-cov` data on top when judging release confidence.
- Mapped suites: 36
- Mapped Rust files: 383
-- Active test blocks: 3782
+- Active test blocks: 3776
- Ignored/manual test blocks: 170
-- Weighted coverage points: 3120.1
+- Weighted coverage points: 3115.0
| Platform | Suites | Active tests | Ignored tests | Weighted points | Layers | Flows | Critical score |
| --- | --- | --- | --- | --- | --- | --- | --- |
-| windows | 32 | 3636 | 156 | 3053.8 | 21 | 11 | 100% |
-| macos | 31 | 3662 | 131 | 3042.3 | 22 | 11 | 100% |
-| linux | 28 | 3219 | 131 | 2677.2 | 20 | 11 | 100% |
+| windows | 32 | 3630 | 156 | 3048.8 | 21 | 11 | 100% |
+| macos | 31 | 3656 | 131 | 3037.3 | 22 | 11 | 100% |
+| linux | 28 | 3213 | 131 | 2672.1 | 20 | 11 | 100% |
+- Mapped Rust files: 382
+- Active test blocks: 3777
+- Ignored/manual test blocks: 170
+- Weighted coverage points: 3115.7
+
+| Platform | Suites | Active tests | Ignored tests | Weighted points | Layers | Flows | Critical score |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| windows | 32 | 3631 | 156 | 3049.4 | 21 | 11 | 100% |
+| macos | 31 | 3657 | 131 | 3037.9 | 22 | 11 | 100% |
+| linux | 28 | 3214 | 131 | 2672.8 | 20 | 11 | 100% |
## Refresh
diff --git a/docs/coverage/CORE.md b/docs/coverage/CORE.md
index 8a1e802d6..d50d6a329 100644
--- a/docs/coverage/CORE.md
+++ b/docs/coverage/CORE.md
@@ -9,10 +9,15 @@ confidence, and criticality.
- Tracked crates: screenpipe-engine, screenpipe-db, screenpipe-sqlite-coordinator, screenpipe-audio, screenpipe-screen, screenpipe-a11y, screenpipe-fs
- Mapped suites: 36
- Mapped Rust files: 383
-- Active test blocks: 3782
+- Active test blocks: 3776
- Ignored/manual test blocks: 170
-- Declared test blocks: 3952
-- Weighted coverage points: 3120.1
+- Declared test blocks: 3946
+- Weighted coverage points: 3115.0
+- Mapped Rust files: 382
+- Active test blocks: 3777
+- Ignored/manual test blocks: 170
+- Declared test blocks: 3947
+- Weighted coverage points: 3115.7
Confidence weights: strong=1.0, partial=0.7, conditional=0.4, smoke=0.3.
Criticality weights: high=1.0, medium=0.7, low=0.4.
@@ -23,16 +28,21 @@ are explicitly enabled in a runtime lane.
| Platform | Suites | Active tests | Ignored tests | Weighted points | Layers | Flows | Critical score |
| --- | --- | --- | --- | --- | --- | --- | --- |
-| windows | 32 | 3636 | 156 | 3053.8 | 21 | 11 | 100% |
-| macos | 31 | 3662 | 131 | 3042.3 | 22 | 11 | 100% |
-| linux | 28 | 3219 | 131 | 2677.2 | 20 | 11 | 100% |
+| windows | 32 | 3630 | 156 | 3048.8 | 21 | 11 | 100% |
+| macos | 31 | 3656 | 131 | 3037.3 | 22 | 11 | 100% |
+| linux | 28 | 3213 | 131 | 2672.1 | 20 | 11 | 100% |
+| windows | 32 | 3631 | 156 | 3049.4 | 21 | 11 | 100% |
+| macos | 31 | 3657 | 131 | 3037.9 | 22 | 11 | 100% |
+| linux | 28 | 3214 | 131 | 2672.8 | 20 | 11 | 100% |
## Crate Summary
| Crate | Suites | Integration files | Source unit files | Active tests | Ignored tests | Weighted points | Flows |
| --- | --- | --- | --- | --- | --- | --- | --- |
-| screenpipe-engine | 12 | 21 | 127 | 1825 | 52 | 1405.2 | 10 |
+| screenpipe-engine | 12 | 21 | 127 | 1820 | 52 | 1401.1 | 10 |
| screenpipe-db | 5 | 65 | 26 | 595 | 26 | 566.6 | 9 |
+| screenpipe-engine | 12 | 21 | 127 | 1821 | 52 | 1401.8 | 10 |
+| screenpipe-db | 5 | 64 | 26 | 594 | 26 | 565.6 | 9 |
| screenpipe-sqlite-coordinator | 1 | 0 | 3 | 27 | 0 | 27.0 | 2 |
| screenpipe-audio | 7 | 26 | 53 | 675 | 49 | 590.3 | 5 |
| screenpipe-screen | 6 | 9 | 19 | 273 | 9 | 243.3 | 4 |
@@ -65,23 +75,38 @@ bun run coverage:core -- --llvm-cov-summary ../../docs/coverage/core-llvm-cov-su
| accessibility | 4 suites / 383 active / 35 ignored / 348.0 pts | 4 suites / 422 active / 10 ignored / 342.9 pts | 4 suites / 341 active / 7 ignored / 313.0 pts |
| audio | 9 suites / 797 active / 50 ignored / 712.3 pts | 8 suites / 791 active / 50 ignored / 708.0 pts | 7 suites / 712 active / 49 ignored / 652.8 pts |
| audio-device | 2 suites / 234 active / 7 ignored / 210.3 pts | 2 suites / 234 active / 7 ignored / 210.3 pts | 1 suites / 155 active / 6 ignored / 155.0 pts |
-| configuration | 2 suites / 153 active / 3 ignored / 138.7 pts | 2 suites / 153 active / 3 ignored / 138.7 pts | 2 suites / 153 active / 3 ignored / 138.7 pts |
+| configuration | 2 suites / 151 active / 3 ignored / 136.7 pts | 2 suites / 151 active / 3 ignored / 136.7 pts | 2 suites / 151 active / 3 ignored / 136.7 pts |
| database | 7 suites / 548 active / 22 ignored / 519.6 pts | 7 suites / 548 active / 22 ignored / 519.6 pts | 7 suites / 548 active / 22 ignored / 519.6 pts |
| db-search | 2 suites / 118 active / 9 ignored / 118.0 pts | 2 suites / 118 active / 9 ignored / 118.0 pts | 2 suites / 118 active / 9 ignored / 118.0 pts |
+| engine-lifecycle | 6 suites / 254 active / 1 ignored / 229.0 pts | 6 suites / 254 active / 1 ignored / 229.0 pts | 6 suites / 250 active / 9 ignored / 228.1 pts |
+| local-api | 3 suites / 476 active / 11 ignored / 336.8 pts | 3 suites / 476 active / 11 ignored / 336.8 pts | 3 suites / 476 active / 11 ignored / 336.8 pts |
+| meeting | 6 suites / 1705 active / 24 ignored / 1390.9 pts | 6 suites / 1705 active / 24 ignored / 1390.9 pts | 4 suites / 1356 active / 20 ignored / 1065.6 pts |
+| ocr | 4 suites / 137 active / 7 ignored / 127.7 pts | 4 suites / 130 active / 7 ignored / 125.5 pts | 3 suites / 121 active / 6 ignored / 116.5 pts |
+| os-integration | 1 suites / 6 active / 0 ignored / 1.7 pts | 1 suites / 6 active / 0 ignored / 1.7 pts | - |
+| performance | 13 suites / 1647 active / 80 ignored / 1457.1 pts | 14 suites / 1755 active / 84 ignored / 1500.3 pts | 13 suites / 1647 active / 80 ignored / 1457.1 pts |
+| pipes | 1 suites / 539 active / 3 ignored / 377.3 pts | 1 suites / 539 active / 3 ignored / 377.3 pts | 1 suites / 539 active / 3 ignored / 377.3 pts |
+| privacy | 5 suites / 1000 active / 42 ignored / 807.4 pts | 5 suites / 1039 active / 17 ignored / 802.3 pts | 5 suites / 958 active / 14 ignored / 772.3 pts |
+| real-app | - | 1 suites / 108 active / 4 ignored / 43.2 pts | - |
+| speaker | 2 suites / 388 active / 11 ignored / 388.0 pts | 2 suites / 388 active / 11 ignored / 388.0 pts | 2 suites / 388 active / 11 ignored / 388.0 pts |
+| storage | 4 suites / 675 active / 39 ignored / 561.0 pts | 4 suites / 675 active / 39 ignored / 561.0 pts | 4 suites / 675 active / 39 ignored / 561.0 pts |
+| sync | 1 suites / 539 active / 3 ignored / 377.3 pts | 1 suites / 539 active / 3 ignored / 377.3 pts | 1 suites / 539 active / 3 ignored / 377.3 pts |
+| configuration | 2 suites / 153 active / 3 ignored / 138.7 pts | 2 suites / 153 active / 3 ignored / 138.7 pts | 2 suites / 153 active / 3 ignored / 138.7 pts |
+| database | 7 suites / 547 active / 22 ignored / 518.6 pts | 7 suites / 547 active / 22 ignored / 518.6 pts | 7 suites / 547 active / 22 ignored / 518.6 pts |
+| db-search | 2 suites / 116 active / 9 ignored / 116.0 pts | 2 suites / 116 active / 9 ignored / 116.0 pts | 2 suites / 116 active / 9 ignored / 116.0 pts |
| engine-lifecycle | 6 suites / 256 active / 1 ignored / 231.0 pts | 6 suites / 256 active / 1 ignored / 231.0 pts | 6 suites / 252 active / 9 ignored / 230.1 pts |
-| local-api | 3 suites / 478 active / 11 ignored / 338.2 pts | 3 suites / 478 active / 11 ignored / 338.2 pts | 3 suites / 478 active / 11 ignored / 338.2 pts |
-| meeting | 6 suites / 1708 active / 24 ignored / 1393.0 pts | 6 suites / 1708 active / 24 ignored / 1393.0 pts | 4 suites / 1359 active / 20 ignored / 1067.7 pts |
+| local-api | 3 suites / 474 active / 11 ignored / 334.8 pts | 3 suites / 474 active / 11 ignored / 334.8 pts | 3 suites / 474 active / 11 ignored / 334.8 pts |
+| meeting | 6 suites / 1706 active / 24 ignored / 1391.6 pts | 6 suites / 1706 active / 24 ignored / 1391.6 pts | 4 suites / 1357 active / 20 ignored / 1066.3 pts |
| ocr | 4 suites / 137 active / 7 ignored / 127.7 pts | 4 suites / 130 active / 7 ignored / 125.5 pts | 3 suites / 121 active / 6 ignored / 116.5 pts |
| os-integration | 1 suites / 6 active / 0 ignored / 1.7 pts | 1 suites / 6 active / 0 ignored / 1.7 pts | - |
-| performance | 13 suites / 1650 active / 80 ignored / 1460.1 pts | 14 suites / 1758 active / 84 ignored / 1503.3 pts | 13 suites / 1650 active / 80 ignored / 1460.1 pts |
+| performance | 13 suites / 1649 active / 80 ignored / 1459.1 pts | 14 suites / 1757 active / 84 ignored / 1502.3 pts | 13 suites / 1649 active / 80 ignored / 1459.1 pts |
| pipes | 1 suites / 540 active / 3 ignored / 378.0 pts | 1 suites / 540 active / 3 ignored / 378.0 pts | 1 suites / 540 active / 3 ignored / 378.0 pts |
| privacy | 5 suites / 1002 active / 42 ignored / 809.1 pts | 5 suites / 1041 active / 17 ignored / 804.0 pts | 5 suites / 960 active / 14 ignored / 774.0 pts |
| real-app | - | 1 suites / 108 active / 4 ignored / 43.2 pts | - |
| speaker | 2 suites / 388 active / 11 ignored / 388.0 pts | 2 suites / 388 active / 11 ignored / 388.0 pts | 2 suites / 388 active / 11 ignored / 388.0 pts |
-| storage | 4 suites / 675 active / 39 ignored / 561.0 pts | 4 suites / 675 active / 39 ignored / 561.0 pts | 4 suites / 675 active / 39 ignored / 561.0 pts |
+| storage | 4 suites / 674 active / 39 ignored / 560.0 pts | 4 suites / 674 active / 39 ignored / 560.0 pts | 4 suites / 674 active / 39 ignored / 560.0 pts |
| sync | 1 suites / 540 active / 3 ignored / 378.0 pts | 1 suites / 540 active / 3 ignored / 378.0 pts | 1 suites / 540 active / 3 ignored / 378.0 pts |
-| timeline | 4 suites / 1225 active / 45 ignored / 1003.9 pts | 4 suites / 1225 active / 45 ignored / 1003.9 pts | 4 suites / 1225 active / 45 ignored / 1003.9 pts |
-| transcription | 6 suites / 872 active / 48 ignored / 681.6 pts | 5 suites / 866 active / 48 ignored / 677.4 pts | 5 suites / 866 active / 48 ignored / 677.4 pts |
+| timeline | 4 suites / 1222 active / 45 ignored / 1001.5 pts | 4 suites / 1222 active / 45 ignored / 1001.5 pts | 4 suites / 1222 active / 45 ignored / 1001.5 pts |
+| transcription | 6 suites / 870 active / 48 ignored / 680.2 pts | 5 suites / 864 active / 48 ignored / 676.0 pts | 5 suites / 864 active / 48 ignored / 676.0 pts |
| ui-events | 4 suites / 817 active / 34 ignored / 620.0 pts | 3 suites / 748 active / 5 ignored / 571.7 pts | 3 suites / 748 active / 5 ignored / 571.7 pts |
| vision-capture | 6 suites / 573 active / 32 ignored / 451.5 pts | 6 suites / 566 active / 32 ignored / 449.3 pts | 5 suites / 557 active / 31 ignored / 440.3 pts |
@@ -135,13 +160,15 @@ bun run coverage:core -- --llvm-cov-summary ../../docs/coverage/core-llvm-cov-su
| db-runtime-reliability | screenpipe-db | windows, macos, linux | database, performance | performance-liveness | high | partial | mixed | 16 | 47 | 6 | SQLite hard-fault classification, failpoint VFS injection, fresh-identity recovery verification with integrity/FK/write canaries, query cancellation, close-severs-connections regressions, multi-pool WAL parity, runtime version pinning, and WAL chaos plus memory-pressure probes, plus read-only quarantine self-heal verification for transient IOERR faults. |
| db-search-indexing | screenpipe-db | windows, macos, linux | db-search, ocr, accessibility, performance | local-api-search, capture-ocr-pipeline, accessibility-ui-events, performance-liveness | high | strong | mixed | 14 | 106 | 4 | FTS, tokenizer, OCR snapshot search, query planning, ordering, accessibility search, and contention coverage. |
| db-timeline-frames | screenpipe-db | windows, macos, linux | database, timeline, storage, performance | timeline-streaming, performance-liveness | high | strong | mixed | 37 | 295 | 13 | Frame/audio joins, timeline query shape, suggestions frames, write queue, DB primitives (src/db.rs split into src/db/ modules), feedback record upserts, media eviction anti-join regressions, SAF output registry, semantic storage, and timeline performance. |
-| engine-api-routes | screenpipe-engine | windows, macos, linux | local-api, timeline, meeting, transcription | local-api-search, timeline-streaming, meeting-live-notes, audio-record-transcribe | high | partial | mixed | 44 | 431 | 6 | Route/unit coverage for search, health, streaming, meetings, time/timezone, and transcription. Legacy endpoint/websocket tests require local data and remain ignored. |
+| engine-api-routes | screenpipe-engine | windows, macos, linux | local-api, timeline, meeting, transcription | local-api-search, timeline-streaming, meeting-live-notes, audio-record-transcribe | high | partial | mixed | 44 | 429 | 6 | Route/unit coverage for search, health, streaming, meetings, time/timezone, and transcription. Legacy endpoint/websocket tests require local data and remain ignored. |
| engine-capture-timeline | screenpipe-engine | windows, macos, linux | vision-capture, timeline, storage, performance | capture-ocr-pipeline, timeline-streaming, performance-liveness | high | partial | mixed | 25 | 306 | 26 | Covers capture trigger logic, frame/audio linking, hot cache, timeline refresh regressions, fragmented MP4 extraction, and HD-mode control. Several real-data tests are intentionally ignored by default. |
| engine-config-lifecycle | screenpipe-engine | windows, macos, linux | configuration, engine-lifecycle, performance | settings-to-engine-config, engine-health-lifecycle, performance-liveness | high | strong | mixed | 13 | 125 | 1 | Fast logic coverage for the config bridge, health-endpoint identity, tray health debounce, sleep/power policies, and queue backpressure. |
| engine-db-recovery-cli | screenpipe-engine | windows, macos, linux | database, engine-lifecycle | engine-health-lifecycle, performance-liveness | high | strong | unit | 1 | 29 | 0 | Exact DB/WAL/SHM working-copy preservation, rollback on archive failure, and restart repair for crashes during the multi-file generation swap. |
| engine-focus-os | screenpipe-engine | windows, macos | engine-lifecycle, os-integration | engine-health-lifecycle, performance-liveness | medium | conditional | unit | 3 | 6 | 0 | Platform focus-tracker parsing/helpers. These files are cfg-gated and only execute on their target OS. |
| engine-linux-keep-awake | screenpipe-engine | linux | engine-lifecycle | engine-health-lifecycle | high | conditional | mixed | 1 | 2 | 8 | Existing isolated-process missing/stalled bus regressions plus ignored live logind inhibitor tests. Live tests require logind and explicit execution. |
| engine-local-api-search-integration | screenpipe-engine | windows, macos, linux | local-api, db-search | local-api-search | high | strong | integration | 1 | 12 | 5 | Active /search route test builds an audio-disabled router, seeds captured-screen-shaped OCR data into an in-memory DB, and asserts the HTTP response and pagination. |
+| engine-meeting-privacy-sync | screenpipe-engine | windows, macos, linux | meeting, privacy, ui-events, pipes, sync | meeting-live-notes, privacy-and-redaction, accessibility-ui-events, performance-liveness | medium | strong | unit | 32 | 539 | 3 | Unit-heavy coverage for privacy filter policy, capture exclusions, UI recorder safety, pipes/live-view/structured-output helpers, sync helpers, and CLI parsing. Meeting detection moved to the engine-meeting-watcher suite. |
+| engine-local-api-search-integration | screenpipe-engine | windows, macos, linux | local-api, db-search | local-api-search | high | strong | integration | 1 | 10 | 5 | Active /search route test builds an audio-disabled router, seeds captured-screen-shaped OCR data into an in-memory DB, and asserts the HTTP response and pagination. |
| engine-meeting-privacy-sync | screenpipe-engine | windows, macos, linux | meeting, privacy, ui-events, pipes, sync | meeting-live-notes, privacy-and-redaction, accessibility-ui-events, performance-liveness | medium | strong | unit | 32 | 540 | 3 | Unit-heavy coverage for privacy filter policy, capture exclusions, UI recorder safety, pipes/live-view/structured-output helpers, sync helpers, and CLI parsing. Meeting detection moved to the engine-meeting-watcher suite. |
| engine-meeting-watcher | screenpipe-engine | windows, macos | meeting | meeting-live-notes | high | strong | mixed | 10 | 270 | 3 | Successor of src/meeting_detector.rs and src/meeting_telemetry.rs. Audio-process and UI-scan backend state machines, candidate resolution, shared telemetry, and a scored trajectory eval. ui_scan/macos.rs and ui_scan/windows.rs are cfg-gated and only execute on their target OS; both backends are null on Linux. |
| engine-retention-storage | screenpipe-engine | windows, macos, linux | storage, engine-lifecycle, performance | engine-health-lifecycle, performance-liveness | medium | strong | mixed | 5 | 39 | 0 | Local retention deletion (including lean-mode heavy-text stripping), cloud archive watermarking, atomic state-file replacement, and the low-disk capture monitor. |
@@ -413,7 +440,7 @@ bun run coverage:core -- --llvm-cov-summary ../../docs/coverage/core-llvm-cov-su
| engine-meeting-privacy-sync | screenpipe-engine | src/meeting_export.rs | source | 7 | 1 | 8 |
| engine-api-routes | screenpipe-engine | src/meeting_summary/extract.rs | source | 7 | 0 | 7 |
| engine-api-routes | screenpipe-engine | src/meeting_summary/finalizer.rs | source | 9 | 0 | 9 |
-| engine-api-routes | screenpipe-engine | src/meeting_summary/notes.rs | source | 8 | 0 | 8 |
+| engine-api-routes | screenpipe-engine | src/meeting_summary/notes.rs | source | 6 | 0 | 6 |
| engine-meeting-watcher | screenpipe-engine | src/meeting_watcher/audio_process/resolve.rs | source | 6 | 0 | 6 |
| engine-meeting-watcher | screenpipe-engine | src/meeting_watcher/audio_process/state.rs | source | 2 | 0 | 2 |
| engine-meeting-watcher | screenpipe-engine | src/meeting_watcher/audio_process/tests.rs | source | 110 | 0 | 110 |
diff --git a/docs/pr-assets/ai-transcript-titles/README.md b/docs/pr-assets/ai-transcript-titles/README.md
index 2acdd5607..23f2f68fd 100644
--- a/docs/pr-assets/ai-transcript-titles/README.md
+++ b/docs/pr-assets/ai-transcript-titles/README.md
@@ -9,8 +9,8 @@ Base: `adeac7f5b` (the list component is unchanged). All captures use a 1280 ×
| before.png | Recreated completed-summary baseline: app-name and missing titles remain generic. |
| after.png | Proposed saved records: two topic titles, a preserved descriptive title, and an unchanged unsummarized row. |
| idle.png | No summary yet; original meeting title. |
-| working.png | Topic heading streams in the summary; original meeting title remains until save. |
+| working.png | Summary streams normally; original meeting title remains until save. |
| ready.png | Saved summary and descriptive meeting title. |
| failed.png | Failed generation; original title remains and retry is available. |
-Persistence and recovery are checked separately by Rust regression tests. The previews illustrate those resulting record states; they do not call the Rust API or a model.
+The previews illustrate intended results of the prompt instructions; they do not call the Rust API or a model. The existing API saves the title supplied by the agent. There is no added title classifier, database guard, special summary heading or title-recovery code. Tests cover installing the revised prompt without changing user settings; live agent compliance has not been tested.
diff --git a/docs/pr-assets/ai-transcript-titles/ready.png b/docs/pr-assets/ai-transcript-titles/ready.png
index 0db0724a1..ca22829ce 100644
Binary files a/docs/pr-assets/ai-transcript-titles/ready.png and b/docs/pr-assets/ai-transcript-titles/ready.png differ
diff --git a/docs/pr-assets/ai-transcript-titles/working.png b/docs/pr-assets/ai-transcript-titles/working.png
index 5dcd2147b..a717be0b9 100644
Binary files a/docs/pr-assets/ai-transcript-titles/working.png and b/docs/pr-assets/ai-transcript-titles/working.png differ