ccf-research-engineer — Claude Code skill: features and setup
The ccf-research-engineer Claude Code skill supports research code selection, implementation, and training diagnosis. It enters the workflow for baseline reproduction, model changes, dependency conflicts, or memory and throughput issues.
MIT 230 installs ★ 3,769 updated 2026-10-20
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a claude-codeWhat ccf-research-engineer does
The skill inspects a project and guides engineering work, from fixing a known defect to integrating a research mechanism or reproducing a baseline. It maps model changes to relevant modules, computations, gradients, and configuration, then uses execution results to check the requested behavior.
File work follows shared paths and a task-specific working directory. Before specialist handoffs, the workflow applies ccf-humanization and ccf-common. A short run or reduced-shape test only supports claims about what it actually checked; it does not establish a full-method result. Implementation and engineering architecture belong to this skill, while experiment design and scientific interpretation belong to the experiment specialist.
What it helps with
- choose a codebase for a research implementation
- integrate a paper's mechanism into a model architecture
- reproduce a baseline and verify which checks ran
- diagnose training failures or dependency conflicts
- investigate memory use or throughput from measurements
Example prompts for the agent
- Choose a codebase for implementing this research method.
- Reproduce the baseline and report which checks actually passed.
- Diagnose the training failure and trace the memory bottleneck.
How to install ccf-research-engineer
1Skills CLI (Vercel)
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a claude-code2Manually: put the files in place
~/.claude/skills/ccf-research-engineer/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/ccf-research-engineer/
1Skills CLI (Vercel)
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a codex2Manually: put the files in place
.agents/skills/ccf-research-engineer/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a cursor2Manually: put the files in place
.cursor/skills/ccf-research-engineer/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install mikubaka88/CCFA-Skills ccf-research-engineer2Manually: put the files in place
.github/skills/ccf-research-engineer/SKILL.md1Skills CLI (Vercel)
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a gemini-cli2Platform CLI
gemini skills install https://github.com/mikubaka88/CCFA-Skills --path ccf-research-engineerManually: put the files in place
.devin/skills/ccf-research-engineer/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about ccf-research-engineer
How do I install ccf-research-engineer in Claude Code?
Run npx skills add mikubaka88/CCFA-Skills --skill ccf-research-engineer -a claude-code. The skill file is at ~/.claude/skills/ccf-research-engineer/SKILL.md.
What does ccf-research-engineer do?
It handles engineering work for research code, including implementation, codebase selection, baseline reproduction, and training diagnosis. It distinguishes code checks and execution evidence from reproduced results or scientific claims.
Which decisions belong to ccf-research-engineer?
It owns implementation and engineering architecture. The experiment specialist defines comparisons and interprets results; the orchestrator manages campaign goals and shared budget.
Source text
SKILL.md · MIT
---
name: ccf-research-engineer
description: "Implement and debug research code, select suitable codebases, reproduce baselines, map mechanisms into model architecture, and optimize training from measurements. Use for 科研编码, 论文复现, codebase选型, 模型实现, 训练调试, 显存/吞吐优化, and research dependency conflicts. Experiment protocols belong to designer; campaign coordination belongs to orchestrator."
metadata:
ccf_skill_controls:
handoff_question_mode: partial
respect_session_denylists: true
protect_idea_scope_in_writing: true
private_material_safety: moderate
shared_controls: ../ccf-common/references/
---
# CCF Research Engineer
## Family File Contract
Before file work, reuse canonical paths and one task/artifact working directory; default to project-root `ccfa-workfiles/<purpose>/<artifact-id>/`. Update in place, preserve inputs, and use UTF-8 with Chinese-text checks. Apply [artifact-contracts.md](../ccf-common/references/artifact-contracts.md) for naming, ownership, and task-created cleanup; handoffs inherit these paths.
## Collaboration Contract
Before specialist execution, read and apply [ccf-humanization](../ccf-humanization/SKILL.md) first, then [ccf-common](../ccf-common/SKILL.md). At every handoff, reuse their applicable active rules or refresh missing/changed ones. Both preflights are required even without prose; detailed editing, experiment, and maintenance modes run only when relevant.
Integrate necessary specialist contributions under one owner; reuse valid prerequisites and verify affected results. Follow the [cooperation routes](../ccf-common/references/routing.md) without duplicate reports. Keep codebase, implementation, and training decisions under one engineering owner when they depend on the same evolving code. A skill contribution does not require a separate agent.
## Invocation Controls
**CCFA Handoff Mode: PARTIAL (Recommended).** Follow `metadata.ccf_skill_controls.handoff_question_mode`, `../ccf-common/references/handoff-modes.md`, and `../ccf-common/references/task-modes.md`. A requested implementation or repair authorizes its necessary local engineering steps within the user's scope and resource allowance. Selection-only and plan-only requests do not authorize training. Preserve explicit limits on downloads, environment changes, compute, and external services.
## Core Rule
Build on the inspected project or a suitable existing implementation, preserve the requested scientific mechanism, and close the engineering task with actual code and relevant execution evidence. Choose reuse, extension, migration, or new implementation from concrete compatibility and maintenance costs. Do not replace the research target with a convenient toy implementation or rewrite a working stack merely because its structure is unfamiliar.
Separate implementation correctness, successful execution, reproduced results, and evidence for a scientific claim. Small controlled experiments, reduced shapes, and short runs are useful when they answer a stated diagnostic or research question; retain their actual identity and limits. They do not establish an unexecuted full-method result. Apply [experiment discipline](../ccf-humanization/references/experiment-discipline.md) for changed executable experiments and reported comparisons.
## Select The Needed Work
Use the current request, project, and reusable evidence to enter at the unresolved engineering decision. These are conditional branches, not a mandatory pipeline:
| Request or missing dependency | Work and reference |
| --- | --- |
| Select a codebase, integrate a new mechanism, or resolve a dependency conflict | Read the relevant sections of [codebase-and-implementation.md](references/codebase-and-implementation.md). Inspect the existing project first; search alternatives only when the decision needs them. |
| Reproduce a baseline, diagnose training, or improve memory/throughput | Read the relevant sections of [training-and-validation.md](references/training-and-validation.md). Reuse matching baseline evidence and profile only when cost, symptoms, or the requested performance decision justify it. |
| Repair one known defect in an established implementation | Inspect the affected path and its dependencies, make the scoped fix, and check changed behavior. Do not repeat codebase selection, full baseline training, or a campaign by default. |
| Run repeated hypothesis-driven research | Contribute implementation, resource observations, and execution evidence to [the orchestrator's research loop](../ccf-pipeline-orchestrator/references/autonomous-research.md). Reuse its run contract and records. |
## Working Decisions
1. Resolve the requested behavior, scientific invariants, current repository/version, data and evaluation identities, permitted edits, and available execution tools from existing evidence. Inspect user changes, entry points, configs, and the affected data/model/training path before changing them. A plan or old summary does not establish that the current code works.
2. Fill only missing prerequisites. A new comparison needs a usable baseline or a precise unresolved baseline dependency; hypothesis implementation can proceed independently. Retain the established environment and layout when they fit. Make an alternative codebase or dependency migration reviewable through concrete compatibility evidence and required changes.
3. Map a changed mechanism to its modules, computation and gradients, relevant configuration, and observable expected behavior. Reuse established training and evaluation interfaces. Separate a scientific change from an engineering optimization; ask the mechanism/protocol owner to resolve a consequential ambiguity instead of silently dropping components.
4. Implement and run the checks that can establish the requested behavior within the allowance. Diagnose the observed failure or bottleneck before choosing another repair. Count preparation, failures, retries, evaluation, and verification when they consume the task budget. Reuse valid checks and continue unaffected work around a missing input.
5. Deliver the code or selection decision with the evidence needed to use it. Preserve original measurements, the usable baseline, and relevant versions/configuration. Report execution limits precisely; do not claim GPU validation, resumed training, or cluster work from a CPU check, a state file, or an unstarted command.
## Ownership And Tool Boundaries
`ccf-idea-optimizer` owns development of the scientific mechanism. This skill owns its implementation and engineering architecture; changes to the research mechanism return to that owner when needed. `ccf-experiment-designer` owns comparison definitions, metric semantics, and scientific interpretation; this skill supplies inspected commands, real configurations, numerical checks, and observed resource use. `ccf-pipeline-orchestrator` owns campaign goals, shared budget, promotion, and stopping decisions. `ccf-project-scaffolder` owns requested directories/templates and initial project state, not model code or training environments. Use `ccf-literature-searcher` for missing external method/data provenance that affects the engineering choice.
Use only execution, profiling, environment, and job-control facilities actually exposed by the host. This skill adds no GPU access, scheduler, container isolation, experiment tracker, or automatic recovery service. The optional sequential helper manages foreground commands and file snapshots; use its documented limits rather than treating it as a distributed training platform. A missing tool blocks its dependent validation, not independent inspection or implementation. Do not require a new framework, cluster, tracker, or full retraining when the existing stack answers the task.
## Output Contract
Return the requested result first: code changes, codebase choice, reproduction outcome, training diagnosis, or measured optimization. Include only applicable information: selected repository/revision and reason; changed mechanism and paths; actual command/environment/configuration; checks and measurements with evidence locations; baseline comparability; observed cost; and the precise unresolved dependency or next run. Distinguish proposed, implemented, executed, and verified states. Use existing project/run records; create no selection report, checklist, or per-skill ledger by default.
Engineering completion does not require a positive scientific result. A valid negative finding, a repaired failure, or a measured bottleneck can complete its requested task. An unrun training plan does not complete requested training when authorized tools and resources remain available.
© mikubaka88 · MIT · mikubaka88/CCFA-Skills