Agent-design-language use-case-writer

Turn one declared source brief, issue, feature, demo, or milestone context into grounded use-case packets with actors, goals, scenarios, success and failure flows, acceptance hooks, assumptions, unsupported assumptions, and non-goals without inventing product commitments or implementation status.

install
source · Clone the upstream repo
git clone https://github.com/danielbaustin/agent-design-language
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/danielbaustin/agent-design-language "$T" && mkdir -p ~/.claude/skills && cp -r "$T/adl/tools/skills/use-case-writer" ~/.claude/skills/danielbaustin-agent-design-language-use-case-writer && rm -rf "$T"
manifest: adl/tools/skills/use-case-writer/SKILL.md
source content

Use Case Writer

Use this skill when a product, runtime, demo, feature, or milestone idea needs a reviewable use-case packet. The skill writes source-grounded use cases; it does not invent roadmap commitments, implementation status, or requirements that are not supported by the supplied brief.

Quick Start

  1. Confirm the declared source:
    • issue body
    • feature brief
    • product note
    • demo plan
    • milestone planning doc
  2. Identify the actors, user goals, system behavior, acceptance hooks, and non-goals that are explicitly supported.
  3. Run the deterministic helper when local filesystem access is available:
    • scripts/write_use_cases.py <use-case-root> --out <artifact-root>
  4. Review unsupported assumptions before treating the packet as product truth.
  5. Stop before creating issues, PRs, implementation plans, or external claims.

Required Inputs

At minimum, gather:

  • mode
  • source
  • audience
  • policy

Supported modes:

  • write_prd_use_cases
  • write_issue_use_cases
  • write_demo_use_cases
  • write_milestone_use_cases

Useful policy fields:

  • source_required
  • unsupported_assumptions_policy
  • implementation_status_claims_allowed
  • issue_creation_allowed
  • write_use_case_artifact
  • stop_before_commitment

If there is no declared source brief or issue context, stop and report

not_run
. Do not backfill missing product intent from plausible-sounding requirements.

Use Case Rules

Each use case should distinguish:

  • actor
  • user goal
  • system behavior
  • trigger
  • preconditions
  • success flow
  • failure or edge flow
  • acceptance hooks
  • evidence source
  • non-goals
  • unsupported assumptions

When source evidence is incomplete, record the unsupported assumption instead of silently turning it into a requirement.

Output

Write Markdown and JSON artifacts when an output root is available.

Default artifact root:

.adl/reviews/use-case-writer/<run_id>/

Required artifacts:

  • use_case_packet.md
  • use_case_packet.json

Use the detailed contract in

references/output-contract.md
.

Stop Boundary

This skill must not:

  • create GitHub issues, PRs, commits, release notes, or implementation plans
  • claim implementation status unless the source explicitly proves it
  • invent product requirements, user personas, business commitments, timelines, success metrics, or demos
  • convert unsupported assumptions into acceptance criteria
  • replace implementation planning, product approval, or milestone approval
  • publish externally without explicit approval

Handoff candidates:

  • finding-to-issue-planner
    when human-approved issue candidates are needed.
  • documentation-specialist
    or docs authoring workflows when a packet should become published documentation.
  • gap-analysis
    when a use-case packet must be checked against implementation or milestone evidence.
  • refactoring-helper
    only when use cases expose code-structure follow-up that needs bounded refactor planning.

Blocked States

Return

not_run
when the source brief or issue context is missing.

Return

blocked
when the requested output requires unsupported product commitments, implementation claims, issue creation, PR creation, publication, or repository mutation.

Return

partial
when a packet can be produced but important actor, behavior, or acceptance evidence is missing.