| name | new-project |
| description | Full research pipeline from idea to paper for the Codex-native Clo-Author. Use when starting a new empirical research project from scratch or bootstrapping this template into a live project. |
New Project
Launch the full research pipeline from idea to paper, orchestrated through the dependency graph.
Read First
AGENTS.md
MEMORY.md
plugins/clo-author/references/runtime-activation.md
plugins/clo-author/references/codex-workflow.md
plugins/clo-author/references/agent-catalog.md
plugins/clo-author/references/critic-runtime-contract.md
plugins/clo-author/references/domain-profile.md
plugins/clo-author/references/logging.md
Pipeline Overview
This skill orchestrates the full dependency graph. Each phase activates when its dependencies are met. The active workflow manages worker dispatch, three-strikes escalation, and quality gates.
Phase 1: Discovery
-> discover interview -> research spec + domain profile calibration
-> discover literature -> literature synthesis + frontier map
-> discover data -> data assessment
Phase 2: Strategy (depends on Discovery inputs)
-> strategize -> strategy memo + strategist-critic review
Phase 3: Execution (depends on approved strategy)
-> analyze -> Data-engineer + data-engineer-critic, then Coder + coder-critic
-> write -> manuscript sections + writer-critic review
Phase 4: Review (depends on approved paper + approved code)
-> review full -> weighted quality score
-> review peer -> editor + blind referees + editorial synthesis
Phase 5: Submission (depends on final gate readiness)
-> submit target / package / audit / final
Workflow
Step 0: Bootstrap and Plan
- Confirm the runtime activation contract before assuming the repo-local skill surface is loaded.
- Clarify the topic, field, target contribution, target journals, and any current materials.
- If the repo is still in template mode, update the project placeholders only after the user confirms the direction.
- Create a requirements spec when the request is broad or ambiguous.
- Save a plan to
quality_reports/plans/YYYY-MM-DD_new-project.md.
- Save an initial session log because this workflow spans multiple phases.
Step 1: Discovery Phase
- If no research spec exists, start with a discovery interview.
- Run the literature workflow:
- Librarian collects the literature
- librarian-critic checks coverage, recency, and frontier gaps
- Run the data workflow:
- Explorer inventories viable datasets
- explorer-critic checks feasibility, measurement, and identification fit
- Save the discovery outputs in
quality_reports/.
Gate: a usable research spec plus at least one of literature review or data assessment must exist before advancing to strategy.
Step 2: Strategy Phase
- Run the strategize workflow.
- Require a saved strategy memo plus a saved strategist-critic review.
- Do not treat the design as approved until the strategist-critic score clears the active gate.
Gate: the strategy memo must pass strategist-critic review with score >= 80.
Step 3: Execution Phase
- Run analyze from the approved strategy artefact.
- Run write only after the code/output side has produced a usable results summary.
- Save scripts, figures, tables, and manuscript artefacts in the standard paths.
Gate: data-preparation work must pass data-engineer-critic review before it becomes approved pipeline input. Estimation code must pass coder-critic review before it becomes approved analysis. Paper sections must be backed by approved results and evidence.
Step 4: Review Phase
- Run a comprehensive review for the current artefacts when the paper is ready for integrated quality control.
- Run peer review when the user wants a realistic editorial/referee simulation.
- Save every review artefact and update the research journal with score, approval state, and next step.
Gate: the project is commit-ready at >= 80, PR-ready at >= 90, and submission-ready only under the final gate.
Step 5: Submission Phase
- Run submit target for journal recommendations.
- Run submit package to assemble the replication package.
- Run submit audit or submit final for the final gate.
User Interaction Points
Pause for user review at the major decision boundaries:
- after the discovery interview if it materially defines the project
- after the strategy memo if multiple credible designs remain
- after the first approved analysis results
- after peer-review synthesis when revisions are required
- before final journal choice or submission packaging
Between those boundaries, the workflow should proceed autonomously through the active worker-critic gates.
Outputs
quality_reports/specs/YYYY-MM-DD_*.md
quality_reports/plans/YYYY-MM-DD_*.md
quality_reports/session_logs/YYYY-MM-DD_*.md
quality_reports/research_journal.md when multi-role coordination is active
- Updated
AGENTS.md and plugins/clo-author/references/domain-profile.md when the template is being converted into a real project