The terminal's search lists the explorer's rows for any explore word (owner:, updated:, archived:), as /explore does #7930

The terminal now treats a search as the explorer's whenever it has any of the five explore words: kind:, sort:, updated:, owner: or archived:. Before, only kind: and sort: did, so owner:trevato alone filtered the board instead of listing apps and people. I didn't try it by eye in the terminal.

• Shared predicate: src/actions/explore.ts exports isExploreLine(line), true when any word starts with one of those five keys followed by :. It matches the case-insensitive, key-with-colon test the dashboard's old exploring regex used, with the three missing keys added.
• Terminal: src/tui/screens/dashboard.ts calls isExploreLine in place of its own regex.
• MCP: explore's doc and its query description in src/http/mcp/tools/explore.ts now name kind:, sort:, updated:, owner: and archived:.
• Tests:

- test/explore.test.ts covers the predicate for each key (including an upper-case key) and for plain words (false). - test/tui-dashboard-explore.test.ts has owner:alice typed alone, giving "Read as: owner:alice" with alice's apps and none of bob's. - I did not run either test against the unchanged code to confirm they fail there.

• Checks: explore.test.ts, tui-dashboard-explore.test.ts, mcp-size.test.ts and bun x tsc --noEmit pass, and bun scripts/gate.ts printed nothing.

A board search that contains owner: text now goes to the explorer, which matches how /explore reads it. I didn't touch the parsing in readExploreLine or any MCP ceiling. The only extra step to try it is typing /owner:trevato in the terminal.

Closes #7926

The terminal now treats a search as the explorer's whenever it has any of the five explore words: kind:, sort:, updated:, owner: or archived:. Before, only kind: and sort: did, so owner:trevato alone filtered the board instead of listing apps and people. I didn't try it by eye in the terminal.

• Shared predicate: src/actions/explore.ts exports isExploreLine(line), true when any word starts with one of those five keys followed by :. It matches the case-insensitive, key-with-colon test the dashboard's old exploring regex used, with the three missing keys added.
• Terminal: src/tui/screens/dashboard.ts calls isExploreLine in place of its own regex.
• MCP: explore's doc and its query description in src/http/mcp/tools/explore.ts now name kind:, sort:, updated:, owner: and archived:.
• Tests:

- test/explore.test.ts covers the predicate for each key (including an upper-case key) and for plain words (false). - test/tui-dashboard-explore.test.ts has owner:alice typed alone, giving "Read as: owner:alice" with alice's apps and none of bob's. - I did not run either test against the unchanged code to confirm they fail there.

• Checks: explore.test.ts, tui-dashboard-explore.test.ts, mcp-size.test.ts and bun x tsc --noEmit pass, and bun scripts/gate.ts printed nothing.

A board search that contains owner: text now goes to the explorer, which matches how /explore reads it. I didn't touch the parsing in readExploreLine or any MCP ceiling. The only extra step to try it is typing /owner:trevato in the terminal.

The terminal's search now lists the explorer's rows for owner:, updated: and archived: as well as kind: and sort:, so one explore line means the same on /explore, in the terminal and through MCP. I haven't run it by eye in the terminal.

The check failed because I changed the wording of MCP explore's doc and query description without regenerating the tool table in docs/MCP.md. I ran bun scripts/mcp-tools.ts, committed the regenerated docs/MCP.md, and pushed.

The other three failures were who filed it, in the terminal (#7658) and the two private destinations in the terminal tests. They pass when I run their files (test/tui-egress.test.ts, test/tui-reports.test.ts) on their own. I think they failed from load in the check's 43-shard run, but I haven't confirmed that, so if they fail again on this push I'll look at them properly.

What changed

• isExploreLine(line) in src/actions/explore.ts is true when any word starts with kind:, sort:, updated:, owner: or archived:, in any letter case.
• src/tui/screens/dashboard.ts uses it in place of its own regex, so owner:trevato alone lists trevato's apps and people, ranked as /explore?q=owner:trevato does.
• MCP explore names all five words in its doc and query description.

Tests

• test/explore.test.ts covers the predicate for each key and for plain words, which give false.
• test/tui-dashboard-explore.test.ts shows owner:alice typed alone giving the explorer ("Read as: owner:alice") with alice's apps and none of bob's.
• I didn't run the new tests against the old code, so I haven't confirmed they fail without the change.

Not done: a board search that contains owner: text now goes to the explorer, which matches /explore and is intended. I didn't change readExploreLine's parsing or any MCP size ceiling.

Try it: in the terminal, type /owner:trevato.

The terminal's search, /explore and the MCP explore tool now share one isExploreLine predicate. It treats owner:, updated: and archived: as explore words. docs/MCP.md is regenerated for the tool's new words.

The failed check does not point at this change. Shard 30 was killed after 41 s while running test/unfurl.test.ts. The log reports 0 failures across 10214 tests, and the other 41 shards passed. This pull does not touch unfurl or anything it uses. It edits src/actions/explore.ts, src/http/mcp/tools/explore.ts, src/tui/screens/dashboard.ts, docs/MCP.md and two explore tests. Run alone here, test/unfurl.test.ts passes (12 of 12 in 2.2 s), so I read it as a slow or stalled shard.

I changed no code in response. I pushed an empty commit, c832d771, so the forge runs the check again. If shard 30 is killed again on a different test, that is a problem with the shard and not this pull.

To try it, type owner:, updated: or archived: into the terminal's search, /explore or the MCP explore tool. Each is read as an explore word. test/explore.test.ts and test/tui-dashboard-explore.test.ts cover this.