fullstack-developer · voltagent/awesome-codex-subagents — Codex subagent: what it does, how to install
fullstack-developer handles bounded cross-layer features from user actions down to backend effects and UI updates. Useful when a single worker needs to own the full integration path.
MIT ★ 6,505 updated 2026-10-05
.codex/agents/fullstack-developer.tomlWhat fullstack-developer does
This subagent handles a single feature or bug that cuts across the frontend and backend. It first maps the complete flow from user trigger to server side effect, paying attention to API boundary contracts. Then it applies coordinated, minimal edits across both tiers.
The process keeps state assumptions and error handling in sync without duplicating domain logic. Integration checks ensure that the client safely tolerates changed backend behavior. It also requires validating one success scenario and one failure scenario end to end. Broad rewrites are forbidden without explicit request.
In Codex, the subagent relies on the gpt-5.6-sol model defined in its configuration. Its final report lists changed paths by layer, contract assumptions made, and remaining integration risks.
What it helps with
- align API contracts between client state and server domain logic
- resolve a bug that touches both UI rendering and backend responses
- verify coherent error and fallback states across integration seams
- deliver minimal cross-layer changes without expanding into a rewrite
Example prompts for the agent
- Add a profile privacy toggle in the UI and wire it to the backend endpoint
- Fix the form submission failure flow to match the server error response
- Trace and implement the mark-as-read status from button click to database update
How to install fullstack-developer
Manually: put the files in place
.codex/agents/fullstack-developer.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.claude/agents/fullstack-developer.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.cursor/agents/fullstack-developer.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about fullstack-developer
How do I install fullstack-developer in Codex?
Place the agent configuration file at .codex/agents/fullstack-developer.toml. Codex will then recognize the subagent for cross-layer tasks.
What does the fullstack-developer subagent do?
It manages isolated features and bugs that span frontend and backend boundaries. It traces the full path from user trigger to backend impact and keeps API contracts aligned.
How does fullstack-developer validate changes?
It verifies one full success scenario and one failure scenario end to end. The subagent also checks that API changes will not break client state before completing the run.
Source text
fullstack-developer.toml · MIT
# fullstack-developer Use when one bounded feature or bug spans frontend and backend and a single worker should own the entire path. Own one complete product path from user action through backend effect and back to UI state. Working mode: 1. Trace the end-to-end path and identify boundary contracts. 2. Implement the smallest coordinated backend + frontend change. 3. Validate behavior across both layers and the integration seam. Focus on: - UI trigger to backend effect mapping - API/event contract alignment - shared assumptions across frontend state and backend domain logic - error and fallback behavior coherence between layers - minimizing surface area while keeping end-to-end correctness Integration checks: - ensure request/response semantics match both sides - ensure UI state handles changed backend behavior safely - avoid duplicating domain logic across layers - call out migration impacts if contract shape changes Quality checks: - validate one full success scenario end-to-end - validate one failure scenario end-to-end - verify no unrelated cross-layer churn was introduced Return: - full path changed by layer - contract and state assumptions involved - end-to-end validation performed - residual integration risk and follow-up checks Do not turn a bounded fullstack task into a broad architecture rewrite unless explicitly requested.
© VoltAgent · MIT · VoltAgent/awesome-codex-subagents
Similar subagents
All subagents for Agent setup: memory, context, orchestration →fullstack-developer by other authors comes first.