code-to-prd
Skill1 starsChanged 34 days agoMITView on GitHub
Reverse-engineer a frontend, backend, or mixed codebase into an evidence-backed Product Requirements Document. Use when documenting existing routes, user flows, APIs, data contracts, permissions, background work, and observable behavior without inventing missing product intent.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- AthenDrakomin-hub/agent-skillstheir own version, first seen Oct 11, 2026
- Hassaan146/claude-skillstheir own version, first seen Jul 25, 2026
- mahiunfiltered/claude-code-skillstheir own version, first seen Jul 8, 2026
- Hussain-Elfalah/Agentic-Skillstheir own version, first seen Jun 4, 2026
- lihanglogan/code-to-prdtheir own version, first seen Mar 17, 2026
Used with
Projects with it also use ai-act-readiness, c-level-agents and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
Hussain-ElfalahSkillai-act-readiness
Also in 3 projects that use this
Hussain-ElfalahSkillc-level-agents
Also in 3 projects that use this
MicrosoftSkillinbox-setup
Set up the Inbox agent with recommended VS Code settings and gh CLI configuration
Also in 3 projects that use this
- +9 more from Hussain-Elfalah
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- Hassaan146/claude-skills★ 18same name
- lihanglogan/code-to-prd★ 9same name
- Hussain-Elfalah/Agentic-Skills★ 2same name
- mahiunfiltered/claude-code-skills★ 1same name
- AthenDrakomin-hub/agent-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository27 other files it gives its agents.
- architectSkill
- audit-discoverySkill
- audit-initSkill
- audit-nextSkill
- audit-resolutionSkill
- audit-statusSkill
- audit-triageSkill
- audit-verificationSkill
- codex-ubuntu-serverSkill
- copilot-review-customizerSkill
- create-pluginSkill
- debugSkill
- deep-reviewSkill
- design-promptsSkill
- diagnose-sessionSkill
- feature-scatteringSkill
- human-execution-runbookSkill
- humanizerSkill
- init-repo-instructionsSkill
- inspect-pluginSkill
- install-agent-pluginsSkill
- local-wikiSkill
- recoverSkill
- review-feedbackSkill
- review-repo-instructionsSkill
- uncompromising-handoverSkill
- update-pluginSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Do not look for a development checkout or another globally installed skill.
- Distinctive: line 58
Only when PRD creation is requested.
“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Code to PRD
Reverse-engineer an existing frontend, backend, or mixed codebase into an evidence-backed Product Requirements Document.
Required Reference Routing
Load only the references needed for the current stage, but do not skip the required stage reference:
| Stage | Required reference |
| Framework detection and source inspection | framework patterns |
| Final validation before delivery | quality checklist |
For mixed or monorepo projects, apply the framework reference independently to every detected application/package. Never select one “primary” framework and ignore the others.
Modes and Installed Helpers
Choose the requested mode: analyze only, create PRD, or update existing PRD. An inventory request never authorizes scaffolding. Updating an existing PRD is a source-backed document edit, not a scaffolder rerun; preserve manual work and the existing organization. Generate only the requested output scope.
Helpers require Python 3.10+ and the standard library. Resolve SKILL_DIR to the
actual directory containing this loaded SKILL.md; it is a shell variable you
set from the installed resource path, not a required host environment variable.
Do not look for a development checkout or another globally installed skill.
Workflow
1. Build the inventory
For an inventory-only request, run only the selected analyzer command. Stdout is the default; -o writes only the explicitly chosen inventory file:
python3 -B "$SKILL_DIR/scripts/codebase_analyzer.py" /path/to/project
python3 -B "$SKILL_DIR/scripts/codebase_analyzer.py" /path/to/project -f markdown
# Only when saving the inventory was requested:
python3 -B "$SKILL_DIR/scripts/codebase_analyzer.py" /path/to/project -o analysis.jsonAn invalid/unreadable root returns nonzero with a stderr diagnostic. Evidence paths are relative to the analyzed repository (project.root is .); retain the chosen root in the current working context. Treat the output as a starting inventory, not as product truth. It deliberately separates:
- frontend pages;
- inbound backend endpoints;
- outbound API calls;
- models, schemas, and DTOs;
- enums and constants.
Review every detected source location and resolve duplicate or generated routes in context. Resolve every endpoint marked unresolved_prefix before treating its path as complete. This commonly includes Django include(), DRF router actions, Express router mounts, and runtime-composed prefixes.
2. Create the evidence workspace
Only when PRD creation is requested. The destination must be absent or empty. A populated directory is refused with a nonzero result; no existing files are deleted or replaced. There is no force/merge mode. An interrupted write may leave new partial output: inspect it, do not automatically delete it to rerun.
python3 -B "$SKILL_DIR/scripts/prd_scaffolder.py" analysis.json -o prd/ -n "Project Name"The generated files contain [TBC] evidence gaps. They are intentionally not presented as a completed PRD.
3. Perform targeted source passes
For every page, inspect:
- rendered regions, labels, controls, tables, charts, modals, drawers;
- defaults, validation, conditional visibility, loading, empty, error, and permission states;
- navigation calls, route parameters, shared state, refresh behavior;
- outbound API calls and response mapping;
- i18n labels and enum-to-label mappings.
For every backend endpoint, inspect:
- method and full composed path;
- path/query/header/body contract;
- serializer/DTO/schema validation;
- guards, decorators, middleware, permissions, ownership checks;
- service calls, database reads/writes, transactions, events, jobs, and external calls;
- status codes, error mapping, idempotency, retries, and environment-dependent behavior.
Perform separate inventory passes for:
- middleware and global guards;
- background tasks, queues, cron, and schedulers;
- migrations and database constraints;
- Django admin configuration;
- environment/configuration branches;
- serializers, forms, validators, and generated API schemas.
4. Write in product language
Describe user-visible or externally observable behavior. Keep implementation detail only when it changes product behavior or defines a contract.
Label extracted source facts, inferred product meaning, and unknown behavior separately. A generated route title is a naming suggestion, not a confirmed business requirement. Do not treat static API-call detection as proof of a live integration. Use [TBC] when evidence is incomplete. State what was observed and what remains unresolved. Do not invent business meaning.
5. Validate before delivery
Load and execute quality checklist. Every checked item must have source evidence. Unresolved items remain explicitly marked [TBC]; they must not be silently treated as complete.
Output Structure
prd/
├── README.md
├── pages/
├── endpoints/
└── appendix/
├── api-inventory.md
├── enum-dictionary.md
├── model-dictionary.md
└── page-relationships.mdTool Limits
The analyzer is stdlib-only, package-scoped, and intentionally conservative. Static extraction cannot prove runtime behavior, authorization semantics, generated routes, middleware ordering, or business intent. The manual evidence pass is mandatory.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

