AGENT SKILL

amend-workitems

Analyze user instructions and safely amend an entire work-item tree in the selected tracker. Use when a user supplies corrections or adjustments for an Epic or Feature and expects the Epic, Features, User Stories, and child Tasks to be scanned, placed, reviewed, and updated consistently. This skill gathers and presents a change set before any write, creates a recoverable tree backup, and never changes work items without explicit approval.

License 许可
MIT
最近更新
2026-11-14
Stars
★ 1
Forks
0

01 前言 Front Matter

name
amend-workitems
license
MIT
description
Analyze user instructions and safely amend an entire work-item tree in the selected tracker. Use when a user supplies corrections or adjustments for an Epic or Feature and expects the Epic, Features, User Stories, and child Tasks to be scanned, placed, reviewed, and updated consistently. This skill gathers and presents a change set before any write, creates a recoverable tree backup, and never changes work items without explicit approval.

02 资源文件 5

03 SKILL.md 正文

Amend Work Items

Safely incorporate approved corrections across a complete Epic or Feature tree. The skill is content-only by default: hierarchy, state, assignee, area, iteration, tags, and unrelated text remain unchanged. Read the references below before making placement or persistence decisions.

References:

  • ./references/placement.md — keyword ranking and type/section placement.
  • ./references/backup-and-artifacts path.md — backup, revision, Obsidian, and recovery rules.
  • ./references/approval-and-persistence.md — change-set approval and delegated writers.
  • ../../references/ticket-structure.md — artifacts path structure and section order.
  • ../../references/tracker-contract.md — the tracker gateway: every read and write goes through its tracker_* tools, never a tracker's own tools.
  • ../enrich-work-item/references/ — canonical sections and examples for Epic, Feature, and Story.
  • ../generate-breakdown-work-items/references/ — Implementation Plans and Task child rules.

Workflow

1. Intake through UI

Ask one question at a time through the host UI:

  1. Collect the amendment instructions verbatim. Do not paraphrase away corrections.
  2. Collect the reference Epic or Feature (tracker id or link, artifacts path, or filesystem path).
  3. Confirm scope as the complete descendant tree and the non-destructive, content-only policy.

Do not read or write the tree before the reference is supplied. If the reference is ambiguous, stop and ask for a resolvable id, URL, or path.

2. Resolve, snapshot, and scan

Read ./references/backup-and-ledger.md before resolving the source. Create a backup of the entire tree before semantic analysis. The snapshot must include every Epic, Feature, Story, and Task body; ids, revisions, parent links, child links, attachments, and the related implementation plans in the items' folders. Verify that the backup is readable and complete. A failed or partial backup is a hard stop.

Resolve the reference and enumerate descendants from Epic → Feature → User Story → Task before proposing changes. Read each node and retain its original revision. When a local Artifacts exists, use obsidian search/CLI first for title, id, and extracted keywords; use rg as the fallback or secondary index. Search exact phrases, identifiers, normalized tokens, and stemmed terms from the instructions. Keep unmatched instructions visible instead of silently dropping them.

3. Analyze and offer placement

Split the instructions into atomic amendments. For each amendment, score candidate nodes and canonical sections using exact identifiers, title matches, phrase matches, and type/section compatibility. Read placement.md and the relevant canonical/example reference before deciding.

Present a single-select UI list for each amendment. Put the recommended placement first, include the reason and affected node/section, and make the final option exactly Other so the user can describe an alternative. Do not apply a placement until the user selects it. If a placement would require a new work item, structural relink, deletion, or state transition, flag it as out of scope and request explicit direction.

4. Build and approve one change set

Produce a consolidated preview containing, for every selected placement:

  • work-item id/path, type, and section;
  • before/after content or an exact insertion/replacement description;
  • linked Task and Implementation Plan consequences;
  • preserved fields and unresolved items;
  • backup location and source revisions.

Re-read current revisions immediately before the approval prompt. If any revision changed since the scan, refresh that node and rebuild the affected proposal. Write the change set as the reviewed drafts, show them with harness workflow show --artifact <draft>, and ask harness decision present --gate publish-items with them as --artifacts. Cancellation or rejection ends the run without writes.

5. Apply using existing Agile Backlog Toolkit skills

After approval, delegate type-specific content work to the existing skills rather than inventing a parallel format:

  • Epic, Feature, or User Story prose/section changes → $enrich-work-item with its canonical references and approval gate.
  • Human-facing wording → $generate-plain-language-documentation.
  • An existing Story's plan-backed Task changes → $generate-breakdown-work-items rules; update the saved Implementation Plan before reconciling its Tasks.
  • A missing required Task explicitly approved by the user → $generate-work-item and then link it as a child of the Story.

Apply approved content top-down. Never mutate hierarchy, state, ownership, area, iteration, tags, or unrelated sections. Reconcile the Task child list without deleting existing Tasks: update only affected Task content, add approved missing Tasks, and report stale/unmatched Tasks for review.

Any change to a Story's Task list invalidates its hour estimates. Whenever Tasks are added, removed, retitled, or rescoped, recompute the whole Story's split, write the new figures, and report every one that moved (old → new). Nothing asked for approval, so nothing may move silently. Re-check the total against the assignee's remaining capacity in the active sprint; if it no longer fits, STOP and ask before writing. Full rules in ../generate-breakdown-work-items/references/atomic-tasks.md. Update related Artifacts Implementation Plans when the amendment changes scope, acceptance coverage, or task sequencing. Read back every changed tracker item and artifacts file, verify parent links and revisions, and run the applicable validate-artifact checks. Stop on the first failed write or read-back; leave the backup intact and report the exact recovery target.

6. Report

Report backup location, scanned tree counts, approved placements, changed nodes, recomputed Task hours (old → new), Task-list and Implementation-Plan updates, validation results, and unresolved instructions. State explicitly that no state or hierarchy changes were made when that is true.

Safety invariants

  • Never write before backup, placement selection, consolidated preview, and explicit approval.
  • Never overwrite a newer revision; refresh and re-propose instead.
  • Never delete, relink, or change state without a separate explicit approval.
  • Preserve original instruction text and all unmatched items in the report.
  • If the tracker and the artifacts disagree, stop and present the conflict; do not guess which copy wins.

04 同仓库更多 Skills

Monolith-INC
Monolith-INC/monolithic-dev-harness

validate-artifact

Validate a single agile artifact (Epic, Feature, User Story, or Task) against the toolkit rule set. Use when the user asks to "validate this story/feature/epic", "check this ticket", "is this artifact ready?", or provides a local draft path or provider item and wants a quality report. Accepts a local draft or a tracker work item. Runs all checks non-blocking and emits a terminal report + persisted report. One artifact per invocation.

★ 1
Monolith-INC
Monolith-INC/monolithic-dev-harness

split-story

Lateral story-sizing skill. Takes a single User Story and determines if it should be split, how many sub-stories to create, and what split pattern to use — then drafts sub-stories and hands them off. Use when the user asks "split this story", "is this story too big?", "analyze this story for sizing", or provides an artifacts path / tracker id / file path / raw text and wants a split recommendation. One story per invocation.

★ 1
Monolith-INC
Monolith-INC/monolithic-dev-harness

research-decision

Research a product, market, domain, competitor, user, or technical question in service of a named decision, producing a cited decision brief. Use when planning depends on current evidence rather than more ideation.

★ 1
Monolith-INC
Monolith-INC/monolithic-dev-harness

product-spec

Distill an idea, brief, PRD, UX contract, architecture spine, issue, or mixed notes into a compact canonical product contract with stable capabilities and companions. Use before backlog decomposition when intent is defined; update or validate an existing product spec through the same skill.

★ 1
Monolith-INC
Monolith-INC/monolithic-dev-harness

product-requirements

Create, update, or validate a right-sized Product Requirements Document with stable requirements, explicit scope, user journeys where useful, and measurable outcomes. Use for multi-person or multi-epic alignment; skip it when a concise product spec is enough.

★ 1
Monolith-INC
Monolith-INC/monolithic-dev-harness

product-brief

Create, update, or validate a concise product brief that captures a clear concept, its users, value, scope, and success criteria. Use when the user wants a one- or two-page product narrative before a PRD or product spec.

★ 1