From claude-tasks
Process the queue's inbox/ items into full ready/ briefs. Human-initiated (never automatic). For each item — dedup check, gather context, fill all frontmatter fields, write goal + criteria + context, commit to ready/, and archive the raw file to inbox/.processed/ (never delete — audit trail). Flag if the inbox grows past 20 items.
How this skill is triggered — by the user, by Claude, or both
Slash command
/claude-tasks:enrichThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Turns raw `inbox/` captures into executable `ready/` briefs. **Run on demand only** —
Turns raw inbox/ captures into executable ready/ briefs. Run on demand only —
human-initiated, never automatic.
/enrich or says "enrich the inbox" / "clear the inbox".inbox/ has 5+ unprocessed items and there's a natural gap.For each file in inbox/ (whether raw: true or partial frontmatter):
grep -rl "<keyword>" "$CLAUDE_TASKS_DIR"/{inbox,ready,in-progress,done} 2>/dev/null
If a match exists, fold this capture's context into the existing brief and archive the raw file — don't create a duplicate.
Pull from every source available before enriching: linked repos/PRs/docs, related prior briefs, the current session, anything the capture references. Don't ask the user for context that's already discoverable.
| Field | Guidance |
|---|---|
title | Scannable one-liner, lead with the noun |
goal | End state, not "do X" |
context | Everything a zero-context agent needs (paths, repos, links, why) |
criteria | Checkable conditions — if you can't check it, rewrite it |
importance | 1–4 (default 3 unless clearly urgent/strategic) |
autonomy | Honest: full ONLY if a future agent can complete end-to-end with zero user input |
estimated-effort | xs / s / m / l / xl |
domain | one of the domains in tasks.toml |
tags | free-form labels (e.g. depends:<other-id>) |
requires-local | true if the brief needs local-machine files or creds |
requires-repo | repo slug if the brief needs a repo other than this queue |
Use add_task.py --ready, or write directly to ready/ with status: ready and commit.
mkdir -p "$CLAUDE_TASKS_DIR/inbox/.processed"
mv "$CLAUDE_TASKS_DIR/inbox/<file>.md" "$CLAUDE_TASKS_DIR/inbox/.processed/<file>.md"
Never delete. .processed/ is the audit trail for where briefs originated.
inbox/ items are not briefs — never present them as the actionable queue.ready/ briefs and the .processed/ archives together.Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub kmmwren/claude-tasks --plugin claude-tasks