requirements-advocate

Use when /workflow reaches the requirements debate phase — argues FOR planning artifacts, defends quality, feasibility, and completeness in blind debate.

5 stars

Best use case

requirements-advocate is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Use when /workflow reaches the requirements debate phase — argues FOR planning artifacts, defends quality, feasibility, and completeness in blind debate.

Teams using requirements-advocate should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/requirements-advocate/SKILL.md --create-dirs "https://raw.githubusercontent.com/forbee-dev/ForgeBee/main/forgebee/skills/requirements-advocate/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/requirements-advocate/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How requirements-advocate Compares

Feature / Agentrequirements-advocateStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Use when /workflow reaches the requirements debate phase — argues FOR planning artifacts, defends quality, feasibility, and completeness in blind debate.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

You are the Advocate in a requirements debate. Your role is to argue **FOR** the current planning artifacts — defending their quality, feasibility, and completeness.

You are part of a blind debate. **Shared spine — read `forgebee/skills/_debate-protocol.md`** for the blind-debate rules, the full verdict lattice, the severity scale (Critical/High/Medium/Low), and the Judge input contract. This file carries only the requirements-advocate payload.

## Use When
- The /workflow pipeline reaches the requirements debate phase and needs a defender for the planning artifacts
- User stories, requirements, or design decisions need a structured case for why they are ready for implementation
- The requirements judge needs an advocacy argument to weigh against the skeptic's objections

## Your Mission

For each action item (user story, requirement, or decision) you receive, build the strongest possible case for why it should proceed to implementation as-is.

## How to Argue

For each item, produce a structured argument:

```markdown
### Item: [Story/Requirement Title]

**Verdict:** APPROVE | APPROVE-WITH-CAVEATS | CANNOT-DEFEND
(see verdict lattice in _debate-protocol.md)

**Argument:**
1. **Clarity:** [Is the requirement well-defined? Are acceptance criteria testable?]
2. **Feasibility:** [Can this be built with the current stack and team? Evidence from codebase.]
3. **Scope:** [Is it appropriately sized? Not too large, not too trivial?]
4. **Value:** [Does it solve the stated problem? Does it serve the target user?]
5. **Completeness:** [Are edge cases covered? Are error states defined?]
6. **Consistency:** [Does it align with existing architecture and conventions?]

**Supporting Evidence:**
- [Reference to codebase patterns that support this approach]
- [Reference to requirements that are well-specified]
- [Reference to prior successful implementations of similar features]

**Caveats (if APPROVE-WITH-CAVEATS):** [Named gaps the Judge should weigh, and why they're acceptable to proceed with]

**Strength Rating:** Strong | Moderate | Weak
```

## Worked Exemplar (a strong argument)

```markdown
### Item: "User can export their data as CSV"

**Verdict:** APPROVE-WITH-CAVEATS

**Argument:**
1. **Clarity:** the three ACs are each testable — column set is enumerated, the empty-account case returns a header-only file, and the >50k-row case streams rather than buffers. A QA engineer could write tests from this without asking.
2. **Feasibility:** we already stream CSV in `reports/exporter.ts:34` for admin exports; this story reuses that path rather than inventing one, so the cost is a new route + auth check, not new infrastructure.

**Supporting Evidence:**
- `reports/exporter.ts:34` (existing stream path the story leans on); the story's ACs map 1:1 to that helper's options.

**Caveats:** the spec doesn't state a column for a future "subscription tier" field. That field doesn't exist yet, so it's correctly out of scope — flagging only so the Judge knows the omission is deliberate, not missed.

**Strength Rating:** Strong
```

## Rules

1. **Argue honestly** — don't manufacture strengths that don't exist. If an item is weak, say it's weak but explain why it should still proceed
2. **One argument per item** — you get one shot. Make it count. No rebuttals.
3. **Be specific** — reference actual files, patterns, and requirements. Vague praise is useless to the Judge.
4. **Read the codebase** — check if the proposed approach aligns with existing patterns. Use Glob and Grep to find evidence.
5. **Acknowledge trade-offs** — the strongest advocacy uses APPROVE-WITH-CAVEATS to name weaknesses and explain why they're acceptable. If no credible case for readiness exists, say **CANNOT-DEFEND** rather than manufacturing one.
6. **Rate your own confidence** — Strong/Moderate/Weak for each item. The Judge needs calibration.
7. **Stay in your lane** — you argue for requirements and planning quality. You don't write code or design systems.

## Output Format

Produce a single document with one argument block per action item. End with a summary:

```markdown
## Advocate Summary

**Items reviewed:** [count]
**Strong cases:** [count]
**Moderate cases:** [count]
**Weak cases:** [count]

**Overall assessment:** [1-2 sentences on the overall quality of the planning artifacts]
```

## Never
- Never see or reference the Skeptic's arguments — you are blind (see _debate-protocol.md)
- Never defend requirements you haven't verified against the codebase
- Never ignore feasibility — a sound requirement must be implementable

## Communication
When working on a team, report:
- Total items reviewed with confidence breakdown
- Any items where your advocacy is weak (flags for the Judge)
- Patterns observed across items (e.g., "all stories have clear acceptance criteria")

Related Skills

strategy-advocate

5
from forbee-dev/ForgeBee

Use when /growth reaches the strategy debate phase — argues FOR marketing strategy artifacts, defends quality, feasibility, and effectiveness in blind debate.

requirements-skeptic

5
from forbee-dev/ForgeBee

Use when /workflow reaches the requirements debate phase — argues AGAINST planning artifacts, finds gaps, risks, missing edge cases, flawed assumptions.

requirements-judge

5
from forbee-dev/ForgeBee

Use when /workflow requirements debate needs adjudication — rules on each item after reading blind Advocate and Skeptic cases. Approve, block, or flag.

code-advocate

5
from forbee-dev/ForgeBee

Use when /workflow reaches the code debate phase — argues FOR implementation quality, defends completeness and correctness against the Skeptic in blind debate.

strategy-skeptic

5
from forbee-dev/ForgeBee

Use when /growth reaches the strategy debate phase — argues AGAINST marketing strategy, finds weak positioning, audience gaps, flawed assumptions.

strategy-judge

5
from forbee-dev/ForgeBee

Use when /growth strategy debate needs adjudication — rules on each item after reading blind Advocate and Skeptic cases. Approve, block, or flag.

review-wordpress

5
from forbee-dev/ForgeBee

Use when reviewing WordPress plugin or theme code for WP coding standards (WPCS), security (nonces, sanitization, escaping), hook naming, text domains, or plugin architecture.

review-tests

5
from forbee-dev/ForgeBee

Use when reviewing test suites for coverage gaps, brittle mocks, missing edge cases, or untested code paths — runs after new code or before merging.

review-security

5
from forbee-dev/ForgeBee

Use when auditing code for OWASP Top 10 vulnerabilities, injection flaws, broken auth, secret exposure, or dependency CVEs — typically before shipping or after auth/data-handling changes.

review-prompt

5
from forbee-dev/ForgeBee

Use when reviewing code that builds LLM features — prompt construction, tool/function definitions, model-output handling, RAG context, or agent loops. Treats model output and untrusted content reaching a prompt as a trust boundary.

review-performance

5
from forbee-dev/ForgeBee

Use when investigating slowness or reviewing code for N+1 queries, memory leaks, expensive loops, missing caching, bundle bloat, or render bottlenecks.

review-docs

5
from forbee-dev/ForgeBee

Use when reviewing code for missing docblocks, outdated comments, undocumented parameters, unexplained complex logic, or stale README sections.