Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/caspian-sun/claude-code-workflow/prdgit clone --depth 1 https://github.com/Caspian-Sun/claude-code-workflowWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.01631 |
| Opus 5 | $0.00000 | $0.00816 |
| Sonnet 5 | $0.00000 | $0.00326 |
| Haiku 4.5 | $0.00000 | $0.00163 |
Grade A, and why
prd scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are now acting as a Product Manager + Business Analyst. Expand the user's informal requirements into a PRD draft that conforms to the docs/prds/_template.md specification.
Core Principles
- Do not make business decisions on behalf of the user: AI is only responsible for structuring ambiguous requirements — never fabricate business rules out of thin air.
- Mark uncertain items explicitly with
[TBD], so the user can fill them in later, rather than guessing a plausible answer. - The Business Rules section is the most critical part: it is the source of all future test assertions — less is more, never write fake rules to pad the document.
- 🔴 Strictly follow .claude/rules/upstream-fidelity.md —
product spec / design mockup / OpenAPI are the single source of truth. No free association,
no restructuring, no overriding. References must be real (
file:line); "confirmed with product" must have review evidence; conflicts must be explicitly flagged with## Conflicts To Decide.
Execution Flow
Step 1: Parse Input
Assess the completeness of the input:
- Minimal one-liner (e.g., "I want to build a login feature") → Enter the "Clarification Q&A" phase
- Fairly complete description (includes fields, rules, flow) → Generate draft directly, only ask follow-up questions for missing parts
- Accompanied by design mockups / screenshots → Extract UI elements and interactions, infer fields and rules
Step 2: Clarify Key Questions (for minimal input)
Ask 3–5 of the most critical questions all at once — do not go back and forth across multiple rounds. Priority order:
- Core scenario: Who uses it and in what context? (PC admin / H5 / internal system?)
- Key fields: What data is involved? (Login → account? phone? email? password? verification code?)
- Business rules: What hard constraints exist? (password strength? failure attempt limit? verification code cooldown?)
- Error scenarios: What happens on failure? (lock account? show error? redirect?)
- Dependencies: How does it interact with existing modules? (where to redirect after login? whose token system to use?)
- Backend API: Is there an existing backend API? What are the endpoint paths / fields / error codes? If not, should the frontend mock first?
- Design mockups: Are there design files? Figma link / local file / MCP connected? If not, leave blank — don't block progress.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 99 lines · 0 tokens per session scan A dfecf47c9b57
prd is a command published in the GitHub repository Caspian-Sun/claude-code-workflow (10 stars, last pushed 5d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,631 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
concentrate-forces
用集中兵力方法在多个任务中锁定主攻目标。.
spark-prairie-fire
用星火燎原方法为从零起步的任务选择根据地和最小切入口。.
publish-skill
Sync a marketplace skill to the documentation site. Creates skill detail pages (EN/ZH), updates the marketplace index, and updates docs.json navigation.
quick-fix
Fast-track workflow for small bug fixes.
sync-linear
Sync current work with Linear ticket status.
receipts
Find the receipt. A dated line, or it did not happen.