Autonomous AI sales agent powered by Claude Code or Codex. Finds prospects, writes cold emails, sends campaigns, handles replies, books meetings. No API keys needed — runs on your Claude or Codex subscriptions.
Keep harvey status working with an invalid config; changelog notes
Ethand0e39a0e7103
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 5804da0..ece3ace 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -19,10 +19,13 @@ minor versions can still change behaviour.approval queue, so with `channels.email.require_approval: true` (the default)Harvey now holds campaign deployment and hands replies to a human instead ofsilently bypassing the setting. `harvey run` and `harvey status` say when this- is happening; prospecting and drafting carry on. Set- `require_approval: false` for Instantly autopilot.+ is happening; prospecting and drafting carry on (drafts wait, using model+ quota, until sending is allowed). Set `require_approval: false` for+ Instantly autopilot.- Disable LinkedIn by default in the model, sample config, and product trainer;save the setup wizard's explicit LinkedIn choice when updating a config.+ Answering "no" to LinkedIn when re-running `harvey setup` now turns it off+ (saved credentials stay in `.env`).- Protect the local dashboard with an access token, restrict requests to trustedlocal origins, and save settings with owner-only permissions. Update and restartHarvey, then open the dashboard link printed in your terminal.diff --git a/harvey/cli.py b/harvey/cli.pyindex bcd4423..7caf06b 100644--- a/harvey/cli.py+++ b/harvey/cli.py@@ -209,10 +209,12 @@ def cmd_status(args):by_provider = summary.get("usage_today_by_provider", {})print(f" Claude calls today: {by_provider.get('claude', summary['usage_today'])}")print(f" Codex calls today: {by_provider.get('codex', 0)}")- from harvey.config import ConfigError, load_config, sending_notice+ from harvey.config import load_config, sending_noticetry:notice = sending_notice(load_config())- except ConfigError:+ except Exception:+ # Status must still print when harvey.yaml is broken; that's+ # when people run it. The --check path reports config errors.notice = ""if notice:print(" Sending: HELD")diff --git a/tests/test_security_defaults.py b/tests/test_security_defaults.pyindex 479841d..0ae9008 100644--- a/tests/test_security_defaults.py+++ b/tests/test_security_defaults.py@@ -165,3 +165,26 @@ async def test_training_on_a_local_url_says_why_it_was_refused():crawler = FallbackCrawler()assert await crawler.crawl("http://localhost:3000", max_pages=3) == {}assert crawler.blocked is True+++def test_status_still_works_when_the_config_is_invalid(tmp_path, monkeypatch, capsys):+ # `harvey status` is what people run when something's wrong, so an+ # invalid harvey.yaml must not crash the sending-hold check.+ import harvey.config+ import harvey.state+ from harvey.cli import cmd_status+ from harvey.config import HarveyConfig+ from pydantic import ValidationError++ def invalid():+ HarveyConfig(channels={"email": {"max_daily_sends": "lots"}})++ try:+ invalid()+ except ValidationError:+ pass+ real_state = harvey.state.StateManager+ monkeypatch.setattr(harvey.state, "StateManager", lambda: real_state(str(tmp_path / "s.db")))+ monkeypatch.setattr(harvey.config, "load_config", invalid)+ cmd_status(SimpleNamespace(check=False))+ assert "Prospects:" in capsys.readouterr().out