fix: stop scanner and roadmap audits from nagging consumer repos

[solomonneas]

What and why

Three Brigade defaults bugs, all surfaced by wiring a consumer repo (bootstrap-doctor) into Brigade. Each makes the daily brief of a repo that uses brigade (as a library and work-loop tool) but does not own the brigade CLI report noise it can't act on. All three were filed as friction from that repo; this is the cleanup.

1. Sweep ingest fails on self-importing scanners

brigade work sweep (ingest on by default) failed the whole sweep with import_path is not configured for memory-refresh and handoff-ingest. Those scanners append to the inbox via their own commands (brigade work import memory-refresh, brigade handoff sync-issues) and produce no JSONL import file, so there is nothing for the sweep to ingest. _scanner_validate_import_output now treats an absent import_path as a self-importing scanner and skips it silently instead of erroring. Workaround was --no-ingest; now unnecessary.

2. chat-memory-sweep required unconditionally

chat-memory-sweep sat in SCANNER_REQUIRED_IDS always, so a code repo with every chat surface disabled had to choose between a failing scanner (no producer for chat-memory-sweeps/latest.json) and a permanent scanner_required warning. It is now required only when at least one chat surface is enabled (via chat_cmd.health); repos that genuinely run chat surfaces are still nudged.

3. roadmap audit counts brigade's own CLI against every repo

brigade roadmap audit introspects brigade's own argparse parser, so a three-command consumer repo reported ~573 "undocumented" brigade commands and a stale docs/command-inventory.md it never owned. The brigade-catalog checks (roadmap_cli_command_missing_docs, roadmap_command_inventory_current) now run only when the target owns the brigade CLI (detected via pyproject.toml name = "brigade-cli"). The universal check that flags a bogus brigade command in any repo's docs (roadmap_documented_command_missing_cli) is unchanged.

Note on intent (please sanity-check)

Bug 3 reverses behavior that an existing test asserted: test_roadmap_audit_includes_command_inventory_drift expected the inventory-drift check to fire on an arbitrary target. I read that as testing the mechanism rather than an intentional "nag every consumer repo" policy, and updated it to mark its target as brigade-cli, plus added a test that a plain consumer repo gets no command-catalog checks. If you actually want every repo audited against brigade's command surface, say so and I'll drop bug 3 from this PR.

Testing

• ./scripts/verify passes: ruff lint + format, mypy, version-sync, and the full pytest suite (2085 passed, 3 skipped) at 81.22% coverage.
• New regression tests: self-importing scanner ingests cleanly; scanner_required ignores a disabled chat-memory-sweep with no surface but warns when a surface is enabled; roadmap audit skips the catalog checks for a consumer repo and keeps them for a brigade-cli target.

[coderabbitai]

Important

Review skipped

Auto reviews are limited based on label configuration.

🏷️ Required labels (at least one) (1)

• coderabbit-review

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: Repository: escoffier-labs/coderabbit/.coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 3ef3a9f8-c31b-4d35-8f84-e81a6f89bfd3

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
• Commit unit tests in branch fix/scanner-roadmap-defaults

---

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