Decision dossier · checked 23 Sep 2026
Loopmark Agent: retain only as an exploratory local drafting CLI.
Loopmark is a local Python CLI for a technical solo operator. It is not a team-approval system, managed scheduler, CRM, or production/unattended publishing service.
Decision card
Exploratory local drafting CLI
Keep Loopmark in the catalogue only for a small, local, draft-only experiment with fictional data and social credentials left unset.
- License
- MIT for the checked repository.
- Evidence status
- No hands-on install or test occurred.
- Do not use for
- Team approvals, managed scheduling, CRM work, or unattended publishing.
What is documented, and what is not
Documented. The repository README describes a Python/LangGraph CLI, local JSON records, an OpenAI dependency, optional LangSmith and social credentials, immediate social posting, and a local scheduler. The scheduler source is a local process.
Not established in checked primary material. A versioned release/package tied to source, a private security-reporting route, backup/restore guidance, approval/RBAC, managed retries/alerts, and durable run visibility. This is scoped missing documentation, not proof that an unreviewed deployment cannot provide it.
The main history showed packaging-related work in August 2026, but the checked repository surface listed no Releases or Packages and no PyPI artifact was independently retrieved. Do not treat the README install line as verified package provenance.
Why the label is narrow
Local JSON is persistence, not documented backup or recovery. A dry-run, a console status, or an exception message is not a team approval gate, managed retry policy, alerting service, or durable operational history. The open issues list managed retries/alerts and org/RBAC/post-approval work as planned paid-layer items.
Examined, but not recommended: MarketingOS AI
| Evidence area | Loopmark | MarketingOS AI |
|---|---|---|
| Boundary | Local drafting CLI with optional local scheduler/social paths. | Broader HubSpot-focused marketing workspace; not a drop-in local CLI. |
| Meaningful documented difference | No verified team approval boundary. | README claims review states, approval decisions and run traces; campaign publishing is intentionally disabled. |
| Why not a replacement | — | Its README says live persistence and production HubSpot sync are incomplete; demo run storage is in-memory. The checked repository has four visible July 2026 commits, no listed release/package, and no verified backup/restore, retry/alert policy, or security-reporting route. |
| Repository source | Loopmark repository | MarketingOS AI repository · MIT license |
Decision: MarketingOS AI was examined because its documented approval-first boundary is useful to compare. It is not recommended and is not added to the catalogue. Replacing Loopmark with it would trade a narrow local CLI for a different, broader stack without resolving release or operating evidence gaps.
Draft-only first test
Do not use customer data, CRM exports, real contacts, production credentials, scheduler/cron, social integrations, sending, posting, or CRM mutation. This proposed test was not performed and cannot establish installation time, output quality, reliability, security, provider handling, or production fitness.
Sources and uncertainty
Primary sources: README/repository, LICENSE, scheduler source, main history, and open issues. Read the full decision report for the change audit and exclusion log.