Is FDEOps a fit?
Where customer decisions meet code.
35 task skills for discovery, delivery and handoff. Use one on its own, or keep an ongoing customer record with fde.
What it adds
Engineering work in an existing repository
Continuity across sessions
A local customer record and bounded resume output, with targeted retrieval of older evidence.
Traceable customer updates
Preparation for independent operation
Choose your starting point
A single task
Install the relevant skill and provide the notes or code it needs. An analysis or draft does not require a customer record.
A continuing customer project
Use fde to select tasks and keep the brief, decisions, evidence and next actions connected.
An existing team setup
Keep your repository conventions, tests and approved tools. FDEOps works with those tools. Another skill pack is optional, and overlapping skill names should be selected explicitly in your agent.
What stays with your team
FDEOps does not provide customer credentials, cloud infrastructure or authority to make production changes. It is not a hosted team workspace, automatic source sync or an enterprise access-control system. Its checks do not certify that a system is safe or that a customer accepted the result.
If a small task is already clear and your existing instructions cover it, you may not need additional structure. Use the parts that help with the work you actually have.
What changes by host
- Claude Code plugin
- Coordinator, individual tasks and session hooks. Hooks use the workspace's customer binding.
- Skill-only installation
- Selected skills available through the host. No automatically registered Claude Code session hooks.
- Workspace adapters
- Pointers in supported agent instruction files for ongoing customer work. They do not install a model or guarantee tool access.
- Local model with a compatible agent
- AI-assisted use depends on the model and host's file and command tools. Validate the setup on representative tasks.
- CLI without an AI model
- Local record operations and generated fieldbook views. It does not perform model reasoning.
Try it against a real task
Choose a small task whose result you can inspect. Give FDEOps the same permitted context you would give your agent normally. Check the output against the source notes, tests or receiving team's requirements. Keep the skill if it improves the result or reduces work you would otherwise repeat.