wso2/labs-agentic-engineer · Archived

amend

Use for a scoped change to an existing PRD — adding a feature, adding an actor, or going deeper on a feature. The instruction names the scope; touch nothing outside it.

First seen Aug 6, 2026

Installation

$ npx skills add wso2/labs-agentic-engineer --skill amend

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from wso2/labs-agentic-engineer · top by installs.

npx skills add wso2/labs-agentic-engineer

Browse all from wso2/labs-agentic-engineer

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 18
License LICENSE
Default branch main
Open issues 99
Status Archived

Skill metadata

Parsed from SKILL.md frontmatter.

More metadata
aep
{"kind":"platform","audience":["design"]}

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,902 B
  • docs SUMMARY.md 183 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 1 installs

SKILL.md

Amend

A scoped edit to specs/requirements/prd.md. The instruction names the scope — a feature to add, an actor to add, or a feature to deepen. Scope is the contract: sections the scope doesn't touch stay byte-identical, and story numbers only ever append (they are permanent — designs, criteria, and tasks cite them).

Revising a point the document already carries is the settle skill's job, not this one: that edit propagates by design, and propagation is the opposite of the promise made above.

The PRD's shape is defined by the prd-contract skill — follow it when writing. The grilling skill owns the question mechanics for every branch below.

The document is already there to ask against, so amend starts where start has to arrive: every question can name the line it would change. Write each answer in as it settles rather than banking them to the end, and keep the edit inside the scope — rounds are cheap, a widened edit is not.

Add a feature

Interview for what the feature does, for whom, and any policy it implies — skipping what the instruction already says. Then:

  • append the feature's stories with fresh numbers,
  • write or extend specs/requirements/features/<slug>.md with the depth, and

link it from the story per the contract — a feature that should not ship yet is an Out of Scope line instead,

  • record any new product decisions (org defaults answer silently, as ever).

Done when every new story has a number and an actor the Actors section defines.

Add an actor

Define the actor in Actors (product-level: name + what they can broadly see/do), then add or amend only the stories their arrival implies.

Two entrances reach this branch, and the second is the common one:

  • The user asked for the actor. The instruction is the brief, and the

stories it implies are the ones it names.

  • You noticed one mid-conversation. "Managers approve them" names a

Manager the Actors section has never defined, and their stories are already under discussion. Define the actor in the same turn those stories are written, so every story names an actor the document defines.

Go deeper on a feature

Expand specs/requirements/features/<slug>.md — interview for the missing depth, then write it there. The PRD body gains at most new story lines the depth surfaced, plus the contract's link to the file if the story does not carry one yet; everything else lands in the feature file.

Close

Summarize exactly what changed — new story numbers, sections touched — in a few lines. The rest of the flow (design, build) stays untouched by this skill.