diff-minimizer
Classifies each planned implementation step against the acceptance criteria so scaffolding and over-build are dropped before code is written. Use when a graduated or full lane change sits between acceptance criteria and the solution write-up.
- License 许可
- MIT
- 所属仓库
- Eliyce/paqad-ai
- 最近更新
- 2026-11-13
- Stars
- ★ 8
- Forks
- 0
01 前言 Front Matter
- name
- diff-minimizer
- metadata
- { "cacheable": "false", "model_tier": "reasoning", "paqad-meta": "skill.meta.yaml", "output_format": "markdown" }
- description
- Classifies each planned implementation step against the acceptance criteria so scaffolding and over-build are dropped before code is written. Use when a graduated or full lane change sits between acceptance criteria and the solution write-up.
02 资源文件 7
- 参考文档agents/openai.yaml157 B
- 资产文件assets/classifications.txt255 B
- 资产文件assets/output.template.md672 B
- 参考文档references/classification-guide.md2.1 KB
- 脚本scripts/extract-ac-ids.sh467 B
- 脚本scripts/lint-output.sh1.5 KB
- 参考文档skill.meta.yaml656 B
03 SKILL.md 正文
What It Does
Evaluates a proposed implementation outline step-by-step against the acceptance criteria. Classifies each step as ac-satisfying, necessary-setup, scaffolding, or over-build and recommends which scaffolding and over-build steps to drop before any code is written.
The goal is to move scope-creep detection upstream from post-implementation review into the planning phase.
Use This When
Between acceptance-criteria-gen and the Story Designer's solution write-up, in the graduated and full lanes. Skip in the fast lane: the overhead is not justified for low-complexity work.
Inputs
- Read the acceptance criteria artifact first; identifiers must be
AC-{fr}.{n}so each step's mapped AC is unambiguous. - Read the proposed solution outline (steps, files to modify, abstractions introduced).
- Read canonical module docs at the module paths listed in
module_doc_paths. Patterns declared there are project conventions and cannot be classified asover-build. - Read
references/classification-guide.mdbefore assigning any classification so the rubric stays consistent.
Procedure
- Run
scripts/extract-ac-ids.sh <ac-file>to load the canonical AC id set; cross-check every cited id against this set. - Parse the proposed solution into discrete steps (one new file, one function, or one significant edit each).
- Classify each step using
assets/classifications.txt(ac-satisfying,necessary-setup,scaffolding,over-build). - For
scaffolding/over-buildsteps, propose a leaner alternative or deletion. - List ACs with no satisfying step under
### Open Questions(or write the literalOpen Questions: none). - Format per
assets/output.template.md; validate withscripts/lint-output.sh.
Output Contract
- Return a heading named
Diff Minimization. - Provide a
Step Maptable with columns#,Step,Classification,Mapped AC,Actionwhere:#is the 1-based step index from the proposed solution.Classificationis exactly one ofac-satisfying,necessary-setup,scaffolding,over-build.Mapped ACis the AC id when the step isac-satisfying, the dependent step number whennecessary-setup, and-otherwise.Actioniskeepforac-satisfyingandnecessary-setup, ordrop: <one-line reason>forscaffoldingandover-build.
- Provide a
Recommended Dropssection listing dropped step numbers with a one-line reason each. - Provide a
Necessary Setup (justified)section listing setup steps and the dependent step numbers. - End with
Open Questionslisting AC ids with no satisfying step, orOpen Questions: noneexactly.
See assets/output.template.md for the canonical shape; the lint script enforces it.
Escalate / Stop Conditions
- Ask when a step's classification depends on a product decision (e.g., whether a validation rule is required or polish).
- Warn when removing a
scaffoldingstep would silently remove error handling the spec implied but did not state: flag for the requirement analyst rather than auto-drop. - Do not classify a step as
over-buildif the canonical module docs at any provided path declare that abstraction as project convention. - Do not invent AC ids; cross-check every cited AC id against the acceptance criteria artifact and emit unknown ids under
Open Questions.
Resources
references/classification-guide.mdscripts/extract-ac-ids.shscripts/lint-output.shassets/output.template.mdassets/classifications.txtagents/openai.yaml
04 同仓库更多 Skills
sample-skill
Sample bundled skill fixture
processing-invoices
Turns a supplier invoice into one ledger line. Use when a supplier invoice arrives and needs booking.
processing-invoices
Turns a supplier invoice into one ledger line. Use when a supplier invoice arrives and needs booking.
understand-idea
Chat and Claude Code: Step 1 of the discover journey: clarifies the goal, who benefits, success and constraints into the idea block. Use when the journey directs it to step 1, not directly.
size-work
Chat and Claude Code: Suggest a size for a ticket from comparable finished work, as a labelled suggestion the team still owns. Use it when someone asks "how big is this?", "roughly how much work is ABC-12?", or wants a backlog item sized. It finds the team's sizing scale at first use, picks two to five of the closest finished tickets with links, and gives a size or range with the drivers behind it and a confidence of low, medium or high, lowered and explained when the area's documents are template text. Every result is labelled a suggested size, never a date or a commitment. It writes nothing into the size field unless the user says "set it", and it refuses to overwrite an existing size unless the user asks to replace it. Produces the sizing block of the record.
refine-ticket
Chat and Claude Code: Improve one or more existing tracker tickets without losing the author's intent. Use it when someone wants a ticket read, diagnosed and tidied up ("this ticket is vague", "fix up ABC-12", "split this, it's too big"). It reads each ticket in full and snapshots it with its last-updated time, says what is wrong with evidence, and proposes a field-by-field before and after change set limited to summary, description, labels, parent, links and new child tickets on a split. It writes nothing until an explicit "apply it" bound to that exact change set, re-checks the ticket for conflicting edits just before writing, never replaces rich descriptions with plain text, never deletes or closes a ticket, and reads each ticket back after applying. Produces the existing and changes blocks of the record.