一键在 Manus 中运行任何 Skill

budget-term-sheet

星标20

分支0

更新时间2026年6月11日 07:25

Build a pre-deploy Rockie GPU budget term sheet before any Rocky-originated experiment submit. Trigger words "/budget-term-sheet", "quote the GPU budget", "show me the term sheet", or any workflow that is about to call `/experiment` for GPU / torch / triton / training / weight-download work.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

Rockielab

Rockielab/rockie-claude

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

/budget-term-sheet

Render the user-facing cost term sheet that Rocky must show before a Rocky-originated POST $ROCKIELAB_API_URL/api/jobs/submit call. This skill is pre-submit UX and approval gating only. It does not replace platform-context's credit reservation, day caps, or hard budget enforcement.

When to invoke

Before /experiment submits a GPU / torch / triton / training / weight-download job.
When the user asks for a dry-run quote without launching compute.
When a user changes the approved budget and Rocky needs to re-render the quote.

Hard rules

Rocky-originated GPU submits go through this skill before runtime/submit.py.
Only explicit approval tokens count: approve, approved, or I approve.
A user-modified budget below estimate_cents is non-submittable in this repo.
Non-available market states never submit.
Keep the user-facing provider deidentified: Rockie GPU, plus SKU/count/region/tier.
Do not dispatch real compute here. Quote and approval only.

Procedure

Assemble a job spec with job_shape, gpu_type, gpu_count, region, tier, and either timeout_seconds or wallclock_minutes. Use region: "us" and tier: "spot" unless the user or workflow requires a different visible capacity slice.
Run scripts/quote_term_sheet.py to build the canonical JSON term-sheet object. Prefer live quote endpoints when credentials exist; use --market-json for tests and dry runs; use --allow-heuristic-dry-run only when the user explicitly wants a heuristic quote.
Render the markdown block with scripts/render_term_sheet.py. The block must show: total estimate, stage breakdown, recommended budget, wall-clock, Rockie GPU compute summary, availability/confidence, and the explicit Approve / Modify / Cancel prompt.
Wait for a user reply. Parse it with scripts/parse_approval.py.
Decision handling:
- approve: keep the current budget and mark the term sheet approved.
- modify: re-render with the new user_budget_cents and ask again.
- modify_then_approve: re-render with the new budget; only submit if the modified budget is still >= estimate_cents.
- cancel: abandon the submit.
- clarify: ask again; do not submit.
Hand the final approved JSON artifact to /experiment/runtime/submit.py through --term-sheet-json, passing matching --region and --tier. The submit helper refuses artifacts or CLI requests that omit or change the approved region/tier. Do not shell raw curl POST /api/jobs/submit.

Commands

Quote JSON:

python3 ${SKILL_DIR}/scripts/quote_term_sheet.py \
  --job-spec-json /tmp/job-spec.json \
  > /tmp/term-sheet.json

Render markdown:

python3 ${SKILL_DIR}/scripts/render_term_sheet.py \
  --term-sheet-json /tmp/term-sheet.json

Parse the user's reply:

python3 ${SKILL_DIR}/scripts/parse_approval.py \
  --estimate-cents 20000 \
  --reply "cap it at $250 and I approve"

Canonical object

The term-sheet JSON must carry these fields:

quote_id
job_shape
availability
status_code
status_reason
quote_source
generated_at
compute
stages
estimate_cents
recommended_budget_cents
user_budget_cents
wallclock_minutes
confidence_bucket
confidence_statement
decision

Use references/template.md as the user-facing block shape.

同仓库更多 Skills

同仓库

clean

Rockielab/rockie-claude

Pre-commit anti-slop audit — checks staged/dirty files for code slop (debug artifacts, single-use helpers, dead imports) AND documentation slop (stale claims, broken internal links, new .md files, redundant sections). Writes a sentinel so the pre-commit-gate hook lets the commit through. MUST be invoked before `git commit` in this repo. Triggers when the agent is about to commit, when user says "commit", "clean", "ready to commit", or when the pre-commit-gate hook blocks.

2026-07-1720

find-skills

Rockielab/rockie-claude

Pull expert skills into this harness from the Rockie platform catalog (~300 skills across ML training/inference, biology, chemistry, physics, databases, coding). Use BEFORE writing domain guidance from scratch or fumbling an unfamiliar framework — if the task names a library (vLLM, GRPO, TRL, AlphaFold, DuckDB, LAMMPS), check the catalog first. Browse `rockie skill catalog --search X --json`, pull into `.claude/skills/`, invoke immediately. Silently no-ops when the Rockie CLI is absent or logged out.

2026-07-1720

autoresearch

Rockielab/rockie-claude

Canonical around-the-clock research loop. Defines the agent's outer loop — read taste corpus + queue, pick the next experiment, mutate the explicitly-declared mutation surface, run the experiment under a hard time budget against a frozen metric, score, codify, repeat. Augmented with Karpathy's sharp primitives (frozen metric, time cap, explicit mutation surface) and a sustained-campaign layer (concurrent run/plan/write-up pipeline, verdict protocol, novelty re-verification gate) for multi-day operations.

2026-07-1720

agent-builder

Rockielab/rockie-claude

Build any goal-declared agent end to end. Turns one declared goal into a portable, self-hardening, publishable agent repo on the Claude-Code harness. Declare goal, pick runtime/tools and write policies/hooks/safeguards, scaffold a portable repo (config-over-code split), build, run a fresh no-memory adversarial gauntlet, iterate with the clean-pass counter resetting on any failed round, publish ONLY after passing twice with zero CRITICALs, then export a .af agent-file. Biased to emit agents that contain their own internal adversarial critic loop. Triggers on "build an agent", "make me an agent that", "agent builder", "scaffold an agent", "publish this agent", "export to .af", "/agent-builder".

2026-07-1020

build-agent

Rockielab/rockie-claude

Lab composer command scaffold for `/build-agent <name>`. In Agent Builder A1 this only creates a lab-scoped draft agent record and opens that lab's agent detail route for editing. It does not build, run, evaluate, observe, or deploy agents yet.

2026-07-1020

deploy-model

Rockielab/rockie-claude

Narrow quickstart wrapper for pre-baked Deploy-card requests. Triggers on server-validated structured quickstart deploy prompts such as "Quickstart deploy request: track: deploy model: <id> source: quickstart-picker quickstart_intent_id: <id>" and explicit "deploy this featured model" requests from Rockie's quickstart picker when validation metadata is present.

2026-07-1020

name	budget-term-sheet
description	Build a pre-deploy Rockie GPU budget term sheet before any Rocky-originated experiment submit. Trigger words "/budget-term-sheet", "quote the GPU budget", "show me the term sheet", or any workflow that is about to call `/experiment` for GPU / torch / triton / training / weight-download work.

/budget-term-sheet

When to invoke

Before /experiment submits a GPU / torch / triton / training / weight-download job.
When the user asks for a dry-run quote without launching compute.
When a user changes the approved budget and Rocky needs to re-render the quote.

Hard rules

Rocky-originated GPU submits go through this skill before runtime/submit.py.
Only explicit approval tokens count: approve, approved, or I approve.
A user-modified budget below estimate_cents is non-submittable in this repo.
Non-available market states never submit.
Keep the user-facing provider deidentified: Rockie GPU, plus SKU/count/region/tier.
Do not dispatch real compute here. Quote and approval only.

Procedure

Assemble a job spec with job_shape, gpu_type, gpu_count, region, tier, and either timeout_seconds or wallclock_minutes. Use region: "us" and tier: "spot" unless the user or workflow requires a different visible capacity slice.
Run scripts/quote_term_sheet.py to build the canonical JSON term-sheet object. Prefer live quote endpoints when credentials exist; use --market-json for tests and dry runs; use --allow-heuristic-dry-run only when the user explicitly wants a heuristic quote.
Render the markdown block with scripts/render_term_sheet.py. The block must show: total estimate, stage breakdown, recommended budget, wall-clock, Rockie GPU compute summary, availability/confidence, and the explicit Approve / Modify / Cancel prompt.
Wait for a user reply. Parse it with scripts/parse_approval.py.
Decision handling:
- approve: keep the current budget and mark the term sheet approved.
- modify: re-render with the new user_budget_cents and ask again.
- modify_then_approve: re-render with the new budget; only submit if the modified budget is still >= estimate_cents.
- cancel: abandon the submit.
- clarify: ask again; do not submit.
Hand the final approved JSON artifact to /experiment/runtime/submit.py through --term-sheet-json, passing matching --region and --tier. The submit helper refuses artifacts or CLI requests that omit or change the approved region/tier. Do not shell raw curl POST /api/jobs/submit.

Commands

Quote JSON:

python3 ${SKILL_DIR}/scripts/quote_term_sheet.py \
  --job-spec-json /tmp/job-spec.json \
  > /tmp/term-sheet.json

Render markdown:

python3 ${SKILL_DIR}/scripts/render_term_sheet.py \
  --term-sheet-json /tmp/term-sheet.json

Parse the user's reply:

python3 ${SKILL_DIR}/scripts/parse_approval.py \
  --estimate-cents 20000 \
  --reply "cap it at $250 and I approve"

Canonical object

The term-sheet JSON must carry these fields:

quote_id
job_shape
availability
status_code
status_reason
quote_source
generated_at
compute
stages
estimate_cents
recommended_budget_cents
user_budget_cents
wallclock_minutes
confidence_bucket
confidence_statement
decision

Use references/template.md as the user-facing block shape.