Install
$ agentstack add skill-wenboxia-doc-generator-doc-generator ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Doc Generator
Turn a finalized product idea into development-ready documents.
Quick start
/doc-generator idea.md # Full suite from idea file
/doc-generator idea.md --prd-only # Just the PRD
/doc-generator idea.md --unified # Single combined blueprint.md
/doc-generator --from-scratch # Interactive: ask questions, then generate
Modes
Full suite (default)
Generates 4 documents sequentially. Best for serious projects.
PRD only (--prd-only)
Just the product requirements document. Best for early-stage validation.
Unified (--unified)
Merges tech spec, dev guide, and CLAUDE.md into a single blueprint.md. Best for solo developers who want one reference document instead of four.
From scratch (--from-scratch)
No input file. Asks the user 5-8 structured questions to build context, then generates documents.
Document 1: PRD (prd.md)
Read the template:
cat references/doc-templates.md
Generate a PRD containing:
- Product overview (one-sentence + expanded)
- Target users (primary persona with demographics, pain points, goals)
- Core features table (feature, priority P0/P1/P2, user story, acceptance criteria)
- Information architecture (page hierarchy, navigation, key journeys)
- Non-functional requirements (performance, security, accessibility)
- MVP scope boundary (IN scope vs OUT of scope for v1)
- Success metrics (metric, target, measurement method)
- Multi-agent assessment (single agent vs Claude Code Teams, with rationale)
PRD user validation
After generating the PRD draft, perform a user perspective check:
Adopt the persona of [target user from the PRD]. You are non-technical.
You only care about whether this product solves your problem.
Read the PRD and answer:
1. What does this product do? (one plain sentence)
2. Does it solve a real problem you have?
3. Would you use it? Why or why not?
4. What would make you stop using it?
5. What's missing that you need?
Present the results to the user as "AI simulation results" (clearly labeled — this is not a human decision point). If the simulated user's understanding diverges from the intended product direction, flag it before proceeding.
Document 2: Technical spec (tech-spec.md)
Generate from the finalized PRD:
- Stack selection table (layer, choice, rationale)
- Project directory structure
- Core module design with interfaces
- Agent mode assessment (single vs Teams, agent role assignments if Teams)
- Development phases (phase, tasks, acceptance criteria, estimated effort)
- Skills dispatch table (phase, task, recommended Skill, rationale)
- Risk register (risk, probability H/M/L, impact H/M/L, mitigation)
Document 3: CLAUDE.md
Generate from tech spec:
- Code conventions (naming, formatting, comment standards)
- Git workflow (branch naming, commit message format, PR process)
- File organization rules
- Testing requirements (unit, integration, coverage targets)
- Development constraints and guardrails
- Agent boundaries (if using Teams mode)
Document 4: Dev guide (dev-guide.md)
Generate from all previous documents:
- Step-by-step operator manual organized by development phase
- Each step specifies: what to do, which tool, which Skill to invoke, expected output
- Human decision points marked (where to pause and think, not just execute)
- Acceptance criteria per phase (how to know this step is done)
- Common problems and solutions
- Multi-agent coordination instructions (if applicable)
Unified mode (--unified)
When --unified flag is used, merge Documents 2-4 into a single blueprint.md:
# [Product Name] — Development Blueprint
## Technical decisions
[content from tech-spec.md]
## Development conventions
[content from CLAUDE.md]
## Execution plan
[content from dev-guide.md, organized by phase]
The PRD remains separate (it's the "what", blueprint is the "how").
Output files
Full suite
| File | Purpose | Reader | |------|---------|--------| | prd.md | Product requirements | You + Claude Code | | tech-spec.md | Technical architecture | Claude Code | | CLAUDE.md | Dev conventions | Claude Code | | dev-guide.md | Operator manual | You |
Unified mode
| File | Purpose | Reader | |------|---------|--------| | prd.md | Product requirements | You + Claude Code | | blueprint.md | Combined tech + conventions + guide | You + Claude Code |
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: wenboxia
- Source: wenboxia/doc-generator
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.