Menu
Enhance a backlog task's title, description, and optionally generate a plan using Claude. Invoke as /backlog-enhance-tasks TASK-N or /backlog-enhance-tasks TASK-N --build-plan.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
One skill for a backlog project's persistent memory. It does two jobs — learn (load the project's tasks, plans, docs, and memory into the session) and store (synthesize the project's state into persistent memory entries). When invoked without a mode it auto-picks: learn at the start of a fresh session, store after work has been done this session, and it asks when that's ambiguous.
Interact with the Backlog CLI — create and manage tasks, plans, comments, labels, projects, memory, docs, and attachments in a local SQLite workspace
Goal-driven autonomous workflow on top of the backlog CLI. Two strict modes — PREP (exhaustive intake, classify the goal shape, decompose into checkpoints + Scout/Judge/Worker tasks, seed the backlog project, stop) and RUN (execute the board, one active task at a time, structured receipts as comments, Worker tasks bounded by allowed_files/verify/stop_if, every checkpoint gated by a spec-driven Judge sub-agent, mandatory final audit before the goal can be marked done). Use when the user says "/backlog-goal <goal>", "/backlog-goal prep <goal>", "/backlog-goal run <slug>", "set a goal end-to-end", or wants a resumable, observable loop that takes a stated goal from blank-page to verified completion through the backlog.
Pick up one task from a backlog project and iterate on it until a Judge sub-agent verifies it is genuinely done, then exit. The "loop" is internal to a single task — implement, verify, judge, fix, re-judge — not across multiple tasks. Invoke as `/backlog-loop <project>` to pull the next highest-priority `todo` task, or `/backlog-loop <project> <ref>` to target a specific task. Use when the user wants the agent to clear one item off the backlog with a built-in verification gate, without engaging the full goal-and-checkpoint workflow of /backlog-goal.
| name | backlog-enhance-tasks |
| description | Enhance a backlog task's title, description, and optionally generate a plan using Claude. Invoke as /backlog-enhance-tasks TASK-N or /backlog-enhance-tasks TASK-N --build-plan. |
Enhance a backlog task's title, description, and optionally generate a plan using Claude.
Invoke as /backlog-enhance-tasks TASK-N or /backlog-enhance-tasks TASK-N --build-plan.
Accepts: a task ref (TASK-N or bare integer), optional --build-plan flag.
./backlog --profile default task show <ref> --json
Parse the JSON. Extract: id, seq, title, description, type, priority, project.
Improve the title for clarity and specificity. Rules:
Rewrite the description as structured markdown with these sections (omit sections that don't apply):
## Context
<1-2 sentences on why this matters — what breaks or is missing without it>
## Acceptance criteria
- [ ] <specific, testable criterion>
- [ ] <specific, testable criterion>
- [ ] ...
## Implementation hints
<optional: file paths, function names, API patterns — only if clearly known>
Rules:
./backlog --profile default task update <ref> \
--title "<improved title>" \
--description "<expanded description>" \
--as "ai:claude-sonnet-4-6"
--build-plan flag provided)Generate a concise implementation plan:
## Steps
1. <first concrete action — file, function, what to change>
2. <next action>
...
## Testing
- <how to verify this works>
Then attach it:
./backlog --profile default plan add \
--task <ref> \
--title "Implementation plan" \
--content "<plan markdown>" \
--as "ai:claude-sonnet-4-6"
After writing back, show a summary:
Enhanced TASK-N: <new title>
Title: <old> → <new>
Description: <N lines added>
Plan: <attached v1 | skipped>
--profile default for all backlog CLI calls in this projectai:claude-sonnet-4-6 (or the current model)type, priority, status, or project — only title and description