From taskmd-lite
Executes tasks from taskmd files by splitting into independent workstreams run in parallel via subagents. Manages progress, worklogs, planning, coordination, tests, and integration for maximum concurrency.
npx claudepluginhub driangle/taskmd --plugin taskmd-liteThis skill is limited to using the following tools:
Pick up a task and execute it by splitting the work into independent workstreams that run in parallel via subagents — no CLI required.
Executes tasks by splitting into independent workstreams run in parallel via subagents. Use for maximum concurrency on complex, multi-part tasks.
Executes parallel subagents with Task tool and run_in_background for independent tasks like multi-file changes, dynamic spawning, or concurrent operations.
Dispatches subagents via Claude Code's Agent tool to parallelize 3+ independent tasks across different files/modules with verification commands. Activates on execution-ready plans or keywords like 'use subagents'.
Share bugs, ideas, or general feedback.
Pick up a task and execute it by splitting the work into independent workstreams that run in parallel via subagents — no CLI required.
The user's query is in $ARGUMENTS (a task ID like 077 or a task name/keyword).
Find the task file:
.taskmd.yaml for custom dir (default: tasks) and workflow modeGlob for <task-dir>/**/*$ARGUMENTS*.mdRead the task file with the Read tool to get the full description, subtasks, and acceptance criteria
Mark the task as in-progress:
Edit to change the status to in-progress in the frontmatterStart a worklog entry (if worklogs are enabled):
.taskmd.yaml for worklogs: true — only create worklogs if explicitly enabled<task-dir>/<group>/.worklogs/<ID>.mdPlan and identify workstreams:
EnterPlanMode to design the overall approachLaunch subagents in parallel:
Agent tool to launch one subagent per independent workstreamisolation: "worktree" for subagents that modify files, to avoid conflictsCoordinate and integrate:
- [x]) in the task file using Edit as they are completedWrite a final worklog entry summarizing what was done, which workstreams ran in parallel, decisions made, and any open items
Mark the task as done:
.taskmd.yaml for workflow mode:verify checks: run them (bash via Bash tool, assert via code inspection)Edit to set status: completedEdit to set status: in-review and add the PR URL to pr arrayEach worklog entry uses a timestamp heading followed by free-form notes:
## 2026-02-15T10:30:00Z
Started divide-and-conquer execution of the search feature task.
**Workstreams identified:**
1. Core search implementation (subagent — worktree)
2. Test suite (subagent — worktree)
3. Documentation updates (subagent)
**Completed:**
- [x] All subagents finished successfully
- [x] Merged worktree changes
- [x] Tests passing after integration
**Decisions:** Used full-text search with SQLite rather than Elasticsearch.
See SPEC_REFERENCE.md (in the plugin root) for frontmatter schema, workflow modes, and verify checks.