[robzolkos]

Fixes #943.

What

Suggested follow-up commands now name the active profile with --profile, however this run chose it. With production as the default profile:

$ basecamp --profile staging spotlight 456 --hints --json
  before: basecamp unspotlight 456
  after:  basecamp unspotlight 456 --profile staging

One helper, followUpScope in internal/commands/helpers.go, decides what a follow-up must name so that a fresh process acts on the same profile and account:

• --profile <name> whenever a profile is active, from --profile, BASECAMP_PROFILE, default_profile, the only configured profile, or the interactive picker. The name is shell-quoted.
• --account <id> when the account came from --account, from a URL, or from the account picker, the same rule as before. An account set in the named profile is not repeated.
• Nothing when no profile is configured and the account comes from config or an exported BASECAMP_ACCOUNT_ID. Users without profiles see no change.

It replaces hasPersistentAccount, replyAccountArg and the templates-only templateCommandContextArgs. The callers in comments show, comments thread, spotlight, unspotlight and templates library / copy / copy-status use it. Template follow-ups already named the active profile, so they do not change.

Why

A follow-up copied into a new shell ran under that shell's profile, which can be a different host or account. If recording 456 exists there, unspotlight 456 acts on the wrong item.

Naming the profile only when it came from --profile was not enough. A profile picked from the interactive picker is not picked again. The pasted command asks again, or fails with "multiple profiles configured but none selected" when not interactive. A shell can also export a different BASECAMP_PROFILE from the one that ran the command. The cost of the rule is a redundant --profile when the profile came from default_profile or the environment.

Breadcrumbs that do not use this helper still carry neither flag. This PR does not change them.

Testing

• TestFollowUpNamesTheProfileHoweverItWasChosen in internal/cli runs spotlight 456 --hints --json through the root command against a local server. It covers no profiles, default_profile, BASECAMP_PROFILE and --profile, and checks both the follow-up and the account the request reached.
• TestSpotlightFollowUpCarriesTheProfileAndAccount covers a profile active without --profile (the picker case), an account from the URL with and without a profile, and --account under a profile.
• TestFollowUpScopeNamesWhatAFreshProcessLacks covers each account source with and without an active profile, and shell quoting. The template tests are back to their main version, which sets the active profile without the flag.
• With the helper reading app.Flags.Profile again, every case with a profile active and no flag fails: 2 of 4 root cases, 3 of 5 spotlight cases, the 3 profile assertions in the helper test, and all 4 template tests. The remaining cases give the same answer under both rules.
• Ran the built binaries against local stub servers, with hints enabled in config. The picker run drives the real picker in a pseudo-terminal, with no default profile, and selects staging. Each request reached the server and account of the selected profile.

Run	main	Previous head 7f8c97e	This branch
spotlight 456 (default_profile production)	unspotlight 456	unspotlight 456	unspotlight 456 --profile production
--profile staging spotlight 456	unspotlight 456	… --profile staging	… --profile staging
BASECAMP_PROFILE=staging spotlight 456	unspotlight 456	unspotlight 456	… --profile staging
staging from the picker, spotlight 456	unspotlight 456	unspotlight 456	… --profile staging
--profile bare spotlight <URL in account 33333>	… --account 33333	… --profile bare --account 33333	… --profile bare --account 33333
--profile staging spotlight 456 --account 44444	… --account 44444	… --profile staging --account 44444	… --profile staging --account 44444

• bin/ci passes

[Copilot]

🟡 Changes recommended

Template follow-ups lose interactively selected profiles, causing repeated prompts or failures in noninteractive use.

1 open issue

• Propagate interactively selected profiles to follow-up commands · New

What changed in this PR

This PR centralizes follow-up command scope handling so suggested Basecamp CLI commands retain an explicit --profile alongside process-local account selections.

Changes:

• Adds a shared helper for profile and account flags.
• Applies it to template, spotlight, and comment follow-ups.
• Updates tests for profile propagation, account provenance, and shell quoting.

[!TIP]
If you aren't ready for review, convert to a draft PR.
Click "Convert to draft" or run gh pr ready --undo.
Click "Ready for review" or run gh pr ready to reengage.

File	Description
internal/​commands/​templates.go	Uses shared scope handling for template follow-ups.
internal/​commands/​templates_test.go	Sets explicit profile flags in follow-up tests.
internal/​commands/​spotlight.go	Carries scope into spotlight follow-ups.
internal/​commands/​spotlight_test.go	Tests profile and account propagation.
internal/​commands/​helpers.go	Introduces the shared follow-up scope helper.
internal/​commands/​comment.go	Uses shared scope handling for comment breadcrumbs.
internal/​commands/​comment_test.go	Tests inherited and process-local scope behavior.

_{Review effort: Balanced}

[robzolkos]

Codex review of 7f8c97e (gpt-6.1-sol, run locally; it has no GitHub threads). It raised the same picker-profile regression as the Copilot thread above, and 43ac084 fixes it. Codex also noted that the PR had rewritten the template tests to set Flags.Profile, which removed the only coverage of an active profile without the flag. templates_test.go is back to main's version, so that coverage is back.

The PR body briefly showed #946's description, because two drafts shared one scratch file. It now has this PR's own description, and it says Fixes #943.

[Copilot]

🟢 Approved

The focused refactor preserves existing account behavior and adds profile regression coverage, with no blocking issues identified.

No open issues 🎉

1 resolved since last review

• Propagate interactively selected profiles to follow-up commands

_{Review effort: Balanced}

[robzolkos]

Codex review of 43ac084 (gpt-6.1-sol, run locally). It found no regressions in this PR across the seven migrated call sites. It raised one finding that predates this PR, which I'm declining:

• An account set inline for one command (BASECAMP_ACCOUNT_ID=99999 basecamp spotlight 456 --profile staging) is treated as persistent, so the follow-up omits --account. The process can't tell an inline assignment from an exported BASECAMP_ACCOUNT_ID. Exporting it is how agents normally set their account, so fixing this would add --account to every suggestion in that common setup to cover a rare one-off. The behaviour comes unchanged from hasPersistentAccount.