Today the only way to see plan progress is to open an agent session and run
/planager-status, which costs tokens and time. A planager status CLI command
gives users (and CI, and scripts) the same table for free. It reuses the convert.py plan
model and parsers added in 0.7/0.8, reads whatever is present in .plans/ (markdown, HTML,
and/or plans.db — so it works for every style and even mid-migration), and needs no new
dependencies. This keeps the ethos: the humans get direct visibility into what their
agents are planning without any runtime component.
Chosen over alternatives (plan templates, handoff skill, git-commit linking) because it is the most obvious daily-use gap: progress visibility currently requires an LLM session.
Done: added plan_progress and collect_plans to convert.py and a planager status subcommand (--path, --all) with an aligned table ordered in-progress, blocked, planning, done. Blocked plans print their last note line below the table. 14 new tests (129 total), README section added, version bumped to 0.9.0. Dogfooded on this repo: correctly read the mixed markdown + HTML plans in .plans/ in one table.