feat: vouch serve fails fast with clear hint when no KB is present

[Tet-9]

Problem

vouch serve in a directory without a .vouch/ KB silently starts
the server and then misbehaves — the stdio transport emits confusing
blank-line Invalid JSON noise and the JSONL transport errors on
every request. There is no upfront check and no actionable message.

Fix

Call _load_store() at the top of serve() before any transport is
started. _load_store() already handles the KBNotFoundError case
by printing a clear run vouch init hint and exiting non-zero, so
the fix is a single line.

A valid KB still starts normally — no behavior change for the happy
path.

Tests

test_serve_fails_fast_without_kb — invokes vouch serve --transport jsonl in a temp directory with no .vouch/ KB and asserts exit code
is non-zero and the output contains the vouch init hint.

198 tests pass, 5 deselected.

Fixes #95

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: d36907de-bd77-45d4-b49c-190438188ad0

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[Tet-9]

@plind-junior , review this when you can 🫡

[Tet-9]

@plind-junior , review this if you may.....would really be much obliged

[Tet-9]

@plind-junior , review time 🫡

[plind-junior]

Fix the conflict

[plind-junior]

Review

Summary: The one-liner fix for #95 is correct and minimal — _load_store() called at the top of serve() before any transport starts does exactly what the issue asks, reusing the existing KB-not-found guard. The PR is sound for its stated purpose, with one blocking concern about a dropped safety fixture and one target-branch question.

What works

• src/vouch/cli.py:992 — _load_store() at the top of serve() is the right call. _load_store already catches KBNotFoundError, prints error: … + hint: run \vouch init`…to stderr, thensys.exit(2)`, so the fix is genuinely one line and reuses a proven path.
• tests/test_cli.py:1489–1494 — test_serve_fails_fast_without_kb uses monkeypatch.chdir(tmp_path) to guarantee a KB-free environment, invokes via CliRunner, and asserts exit_code != 0 plus "vouch init" in output or stderr. The or is correct because CliRunner() with default mix_stderr=True folds stderr into result.output, so the check is resilient either way.
• The PR has a single commit, keeping the history clean.

Suggestions

• [blocking] tests/test_cli.py — The diff removes monkeypatch.setenv("VOUCH_AGENT", "human-reviewer") and its explanatory comment from test_crystallize_cli_partial_failures_shows_warning. That fixture was added deliberately (commit 7cc04b6, PR fix: allow crystallize to approve own session proposals #49) to break the self-approval guard collision when the OS login happens to be "a". Dropping it reintroduces environment-dependent flakiness on any machine where os.getlogin() returns "a". This removal appears to be an accidental artifact of the branch base, not intentional. The line should be restored before merge.

• [non-blocking] The PR targets the test branch, not main. The resulting diff carries a large amount of apparent regressions (removal of on_progress callbacks, --json flags on lint/search/doctor, vouch init --template, colorised output, multi-id approve, docs/PROJECT-INDEX.md, tests/test_cli_output.py, tests/test_templates.py, version rollback to 0.0.1). If test is intentionally an earlier baseline these are non-issues, but if the intent is to land this on main/release/0.1.0, the base branch needs to be corrected — otherwise all those features get reverted on merge.

• [non-blocking] tests/test_cli.py:1494 — The assertion "vouch init" in result.output or "vouch init" in (result.stderr or "") is slightly over-specified; with the default CliRunner() (no catch_exceptions=False, mix_stderr defaults to True in Click 8) result.output already contains stderr. A single assert "vouch init" in result.output would be cleaner and avoids the (result.stderr or "") dance. Minor.

Verdict

Request changes — The core fix is correct and the new test covers the happy/sad path well. The blocking item is the dropped monkeypatch.setenv("VOUCH_AGENT", "human-reviewer") in the existing crystallize partial-failure test, which reintroduces a known flakiness vector. Restore that line and confirm the target branch is intentional, then this is good to merge.

[Tet-9]

Review

Summary: The one-liner fix for #95 is correct and minimal — _load_store() called at the top of serve() before any transport starts does exactly what the issue asks, reusing the existing KB-not-found guard. The PR is sound for its stated purpose, with one blocking concern about a dropped safety fixture and one target-branch question.

What works

• src/vouch/cli.py:992 — _load_store() at the top of serve() is the right call. _load_store already catches KBNotFoundError, prints error: … + hint: run \vouch init…to stderr, thensys.exit(2), so the fix is genuinely one line and reuses a proven path.
• tests/test_cli.py:1489–1494 — test_serve_fails_fast_without_kb uses monkeypatch.chdir(tmp_path) to guarantee a KB-free environment, invokes via CliRunner, and asserts exit_code != 0 plus "vouch init" in output or stderr. The or is correct because CliRunner() with default mix_stderr=True folds stderr into result.output, so the check is resilient either way.
• The PR has a single commit, keeping the history clean.

Suggestions

• [blocking] tests/test_cli.py — The diff removes monkeypatch.setenv("VOUCH_AGENT", "human-reviewer") and its explanatory comment from test_crystallize_cli_partial_failures_shows_warning. That fixture was added deliberately (commit 7cc04b6, PR fix: allow crystallize to approve own session proposals #49) to break the self-approval guard collision when the OS login happens to be "a". Dropping it reintroduces environment-dependent flakiness on any machine where os.getlogin() returns "a". This removal appears to be an accidental artifact of the branch base, not intentional. The line should be restored before merge.
• [non-blocking] The PR targets the test branch, not main. The resulting diff carries a large amount of apparent regressions (removal of on_progress callbacks, --json flags on lint/search/doctor, vouch init --template, colorised output, multi-id approve, docs/PROJECT-INDEX.md, tests/test_cli_output.py, tests/test_templates.py, version rollback to 0.0.1). If test is intentionally an earlier baseline these are non-issues, but if the intent is to land this on main/release/0.1.0, the base branch needs to be corrected — otherwise all those features get reverted on merge.
• [non-blocking] tests/test_cli.py:1494 — The assertion "vouch init" in result.output or "vouch init" in (result.stderr or "") is slightly over-specified; with the default CliRunner() (no catch_exceptions=False, mix_stderr defaults to True in Click 8) result.output already contains stderr. A single assert "vouch init" in result.output would be cleaner and avoids the (result.stderr or "") dance. Minor.

Verdict

Request changes — The core fix is correct and the new test covers the happy/sad path well. The blocking item is the dropped monkeypatch.setenv("VOUCH_AGENT", "human-reviewer") in the existing crystallize partial-failure test, which reintroduces a known flakiness vector. Restore that line and confirm the target branch is intentional, then this is good to merge.

All right....I'm on it

[Tet-9]

@plind-junior , you may review this now

[plind-junior]

Thanks for the contribution!
From now on, please focus on the desktop app release