Skip to main content

backlog-manager

Manages an outcome-focused backlog stored in /work/backlog.md. Handles command-driven operations to add, refine, list, start, complete, and archive work items. Each item is tracked with a status, creation date, and a clear observable outcome statement. Use when asked to manage a backlog, add work items, or check what to work on next.

Informations de source

Dépôt
davidamitchell/Skills
Dernière activité de la source
1 mai 2026 à 21:47
Langue détectée de SKILL.md
anglais
Étoiles
0
Forks
1

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
backlog-manager
version
1.0
description
Manages an outcome-focused backlog stored in /work/backlog.md. Handles command-driven operations to add, refine, list, start, complete, and archive work items. Each item is tracked with a status, creation date, and a clear observable outcome statement. Use when asked to manage a backlog, add work items, or check what to work on next.
# Backlog Manager ## When Not to Use - When the task is to execute ready items and perform the actual work they describe — use `backlog-worker` instead - When the project requires complex dependency tracking between items - When multiple people need concurrent write access (this is a single-file, single-user system) - When risk tracking, resource allocation, or sprint planning are required - When the work is exploratory and outcomes cannot yet be stated (use a notes file instead until outcomes can be defined) --- ## Interaction Protocol **On ambiguous commands**, ask before acting: - If `Add` input lacks a clear outcome, ask: "What does success look like for this item?" - If `Refine` input is unclear, ask which specific aspect to improve: Outcome, Context, or Notes **Output style**: - `List` and `Next`: plain text only, no commentary - All other commands: confirm the action taken in one sentence, then show the updated item - Do not volunteer prioritization opinions or suggest what to work on unless `Next` is called --- ## Inputs and Outputs **Input**: A command string (`Add`, `Refine`, `List`, `Next`, `Start`, `Complete`, `Archive`) with an optional item ID and description **Output**: Confirmation of action and/or formatted item(s) from `/work/backlog.md` **Composability**: Use alongside `strategy-author` (to derive outcomes from strategy) or `research` (to inform item context); use `backlog-worker` to execute ready items and advance them to done --- ## 1. Source of Truth All work is stored in `/work/backlog.md`. If the file does not exist, create it. The file is the only source of truth. Read it fresh at the start of every session. No reliance on memory. --- ## 2. File Structure The file is an ordered list of work items using this format: ```md # Backlog --- ## W-0001 status: needing_refinement created: 2026-02-21 updated: 2026-02-21 ### Outcome Clear, observable end state. Must describe what is true when complete. ### Context Optional background or rationale. ### Notes Free-form thinking. --- ## W-0002 status: ready created: 2026-02-22 updated: 2026-02-22 ### Outcome System supports distributed rate limiting across horizontally scaled nodes. ### Context Current limiter is instance-local. ### Notes ``` Rules: - Items appear in creation order. - Order represents priority. - IDs are sequential and never reused. - Items are never deleted. - Items are never silently reordered. --- ## 3. Allowed Status Values - `needing_refinement` - `ready` - `active` - `done` - `archived` No other statuses are permitted. --- ## 4. Outcome Standard Every item must contain a clear Outcome section. Outcome must: - Describe an observable result - Avoid describing tasks or implementation steps - Avoid vague language - Be testable in principle If an added item does not meet this standard, it must default to `status: needing_refinement`. --- ## 5. Commands ### Add ``` Add: <title or description> ``` Action: - Generate next sequential ID - Create new item at end of file - Set `status: needing_refinement` - Set `created` and `updated` to today - Convert input into a clear outcome statement if possible - If clarity is insufficient, preserve the original text under Notes and leave Outcome minimal Do not decompose. Do not infer scope beyond the text. --- ### Refine W-XXXX Action: - Improve Outcome clarity - Remove ambiguity - Move supporting material into Context or Notes - Change status to `ready` only if the outcome is specific and observable - Update `updated` date --- ### List Action: - List all items as: `W-XXXX – status – first line of Outcome` - Preserve file order - No commentary --- ### Next Action: - Return the first item with `status: ready` - If none exist, return the first item with `status: needing_refinement` - No reprioritization --- ### Start W-XXXX Action: - Change status to `active` - Update `updated` date --- ### Complete W-XXXX Action: - Change status to `done` - Update `updated` date --- ### Archive W-XXXX Action: - Change status to `archived` - Update `updated` date --- ## 6. Constraints - No deletion - No reordering - No status changes without an explicit command - No rewriting Outcome unless explicitly refining - No autonomous prioritization - No hidden scoring --- ## 7. Session Initialization On session start: 1. Read `/work/backlog.md` 2. Validate: - Unique IDs - Sequential numbering - Valid statuses - Each item contains an Outcome section 3. Report only if inconsistencies exist
Voir sur GitHub