メニュー
Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | context-init |
| description | Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed. |
Create a new context file for a research Phase. One file per Phase — subsequent checkpoints append to this file.
Once at the beginning of each research Phase, imported by the plan file:
step: "import context-management:context-init"
Get the current time:
date +%Y-%m-%d-%H-%M
Output format: yyyy-mm-dd-hh-mm (e.g., 2026-05-16-14-30)
Based on the current Phase context, decide a topic slug:
lit-survey, gap-analysis, protein-folding-ideationCreate the file at: context/<timestamp>-<topic-slug>.md
Write the initial content:
# <Phase Topic>
> Created: <yyyy-mm-dd hh:mm>
> Topic: <Research Topic>
> Phase: <Phase Name>
## Plan Context
<Plan excerpt for this Phase: objectives, expected outputs, constraints, relationship to other Phases in the pipeline>
The Plan Context section should contain the relevant plan excerpt describing this Phase's goals, expected deliverables, and how it connects to the broader research pipeline.
No mid-paragraph line breaks: Write each prose paragraph as a single continuous line. Do not insert newlines inside a paragraph to wrap it at a column width. Newlines are only for separating paragraphs, list items, headings, and fenced code blocks. This applies to the Plan Context written here and to all content appended later by context-checkpoint.
If context/INDEX.md does not exist, create it:
# Context Index
| File | Phase | Topic | Checkpoints | Last Updated |
|------|-------|-------|-------------|--------------|
Add a new row for this file:
| <filename> | <Phase Name> | <Research Topic> | 0 | <yyyy-mm-dd hh:mm> |
If a context file for the current Phase already exists (check INDEX.md for a matching Phase name created in the current session), skip creation and return the existing file path.
The file path of the created (or existing) context file. This path is used by subsequent context-checkpoint calls within the same Phase.
Append research process and results to the current Phase's context file. Each append MUST contain >=500 lines of markdown covering both process and results. Use this skill at plan-designated checkpoint points — typically after each strategy completes or at key decision nodes within a research Phase.
The optimizer brain for the ladder-foundry pretraining loop. Runs the two-level nested batch loop, delegates gating to gate_eval, attributes a failing batch to one weight (attribute-first), and recovers from disk after compaction. Control flow is fully scripted; only the backprop attribution is a judgment call.
Loss-2 judge (codex role). Over one topic's 6 shuffled research-design samples, pairwise-rank by quality using the D1–D5 standard. Emit the pairwise log; the harness computes the order and the ladder verdicts. Judge quality difference, never against academic standards.
Loss-1 judge (codex role). Given one sample's de-identified dialogue and its PolicyCard, decide axis-by-axis whether the user-simulator enacted the card's per-axis pressure. Judge enactment of the card, never whether the research is good.
Closing skill for the research-executor, loaded as the last step of formated-specs. Summarize the design just produced into one research-result JSON fenced block in your reply. Do not execute the research.
Spec-slot skill for the research-executor. Emit the 4-layer DARE orchestration of the assigned topic as one research-graph JSON fenced block in your reply. Replaces the generic spec-writing step.