fix(recall): show auto-recall failures

[noctuid]

Surface auto-recall errors through TUI warning notifications instead of relying on console output that disappears during redraw. Continue to inject no memory context when recall fails.

[Copilot]

🔵 Needs a closer look

Empty-string recall errors may not produce warnings, and failed-context clearing lacks integration coverage.

Pull request overview

This pull request surfaces auto-recall failures as TUI warnings while preventing failed recalls from injecting memory context.

Changes:

• Propagates auto-recall errors.
• Displays failures through warning notifications.
• Updates tests and changelog documentation.

File summaries

File	Summary
tests/recall.test.ts	Updates recall error-result expectations.
tests/bootstrap.test.ts	Tests warning notifications.
src/index.ts	Propagates and displays auto-recall failures.
CHANGELOG.md	Documents the fix.

Review details

Suppressed comments (2)

src/index.ts:663

• This truthiness check drops failures whose message is an empty string: the RecallClient contract permits error: "", and the catch path also returns Error("").message unchanged. In that case the recall is cleared but no warning is shown, contrary to the goal of surfacing every auto-recall failure; discriminate on undefined instead.

    if (result?.error) {

tests/bootstrap.test.ts:3521

• This integration test only verifies the warning. The failure path also clears the module-level lastRecallMessage, which the separate context handler later consumes; without invoking that handler here, a regression could still inject stale or new memory context while this test passes. Please assert that the context result contains no recall message after the failed before_agent_start (ideally after first seeding a successful recall).

    const ctx = createMockContext();
    await pi.handlers.get("before_agent_start")!(
      { type: "before_agent_start", prompt: "Hello" },
      ctx
    );

• Files reviewed: 4/4 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.