Terminal
Terminal/Skill
Plans, runs and monitors model fine-tuning on Fireworks from a coding agent, including cost confirmation, checkpoints, deployment and teardown.
By Fireworks AI
Works with
npx skills add fw-ai/cookbook --skill configureConfigure: plan, run, and monitor Fireworks training.
This skill owns the execution workflow. The installed fireworks-training
compatibility skill carries the detailed references; the public cookbook carries
the executable recipes.
| Signal | Route |
|---|---|
| Task-shaped goal without clear method, data, or evaluation | research |
| Failed, stuck, errored, or unexpectedly low-quality run | debug |
| Explicit train, deploy, resume, or research handoff | stay in configure |
When a research handoff exists, reuse its case_study when present,
cookbook_entry_tier, cookbook_entry_path, notebook, readme,
implied_method, dataset_plan, eval_plan, and suggested_path. Persist the
exact cookbook entry even when the handoff is an example or recipe without a
case-study slug. Do not ask again unless the user corrects it.
Research owns the welcome-entry question and its privacy notice. Route a
vague first message there instead of asking a second welcome question.
Use one UUID for the whole run:
export FIREWORKS_SESSION_ID="$(python3 -c 'import uuid; print(uuid.uuid4())')"
export FIREWORKS_CLIENT_SOURCE="fireworks-training-skill/2.2.0"
Record entry_skill: configure in the private run manifest. Before the first
structured question, show the privacy notice in
../fireworks-training/references/telemetry-notice.md.
Check firectl skill-journey record --help once. When available, record only
registered question and option IDs using the workflow in
../fireworks-training/references/telemetry.md. If unavailable, recommend a
firectl upgrade and continue with local run state. Telemetry must never block
training.
firectl ... --help for commands and flags.Never guess current models, shapes, prices, limits, or defaults.
Before any firectl call, run firectl whoami. If authentication is missing,
tell the user to enter a scoped API key in their terminal with hidden input:
read -s FIREWORKS_API_KEY
echo
export FIREWORKS_API_KEY
firectl whoami
Never ask for an API key in chat or echo it.
Read references/path-intake.md.
question_id: configure-q-path answer; research-q3 never counts.Confirm:
firectl version, authentication, quota, and billing readiness.Do not upload or create resources during preflight.
Read references/cost-estimation.md.
Cost estimation does not replace the mandatory final-plan confirmation.
Before any upload, evaluator registration, paid inference, training creation, checkpoint promotion, or deployment, show one complete plan:
platform-resolved, unknown before create; never guess or stall the plan.Ask the user to approve that exact plan. Any change to method, model, parameters, sweep breadth, or cost ceiling requires renewed confirmation. Promotion and deployment require separate confirmation.
Read-only get, list, whoami, quota, and catalog commands are allowed.
After approval, attempt the documented mutation. If firectl returns
BLOCKED: mutating command, show the exact reconstructed command and ask the
user to run it. Never unset agent-detection variables, configure a safe-account
override, or switch tools to bypass the guard.
The guard can also block a mutating command's dry-run form. In that case the user runs the dry-run and returns its output before final approval.
fireworks-training-runs/<run-id>/run.md.AlreadyExists, query the planned ID and
reuse only an exact configuration match.debug. Never launch a replacement before reconciling the prior run.For Training API SFT and RL, explicitly pass emit_grad_norm_metrics=True to
each optim_step and log result.metrics with the optimizer step. Telemetry
defaults to off; verify it appears after the first step. For async_rl_loop,
set grad_norm_metrics="basic" (or "detailed"). See
gradient norm metrics.
| Need | Reference |
|---|---|
| Path intake | references/path-intake.md |
| Cost contract | references/cost-estimation.md |
| Configure response shapes | references/output-template.md |
| Installation and authentication | ../fireworks-training/references/getting-started.md |
| Method and data selection | ../fireworks-training/references/choose-method.md |
| Preference data and evaluators | ../fireworks-training/references/preference-data-and-evaluators.md |
| Managed RFT (deprecated; existing jobs only) | ../fireworks-training/references/managed-rft-operations.md |
| RFT tracing (deprecated; existing jobs only) | ../fireworks-training/references/rft-agent-tracing.md |
| Training API | ../fireworks-training/references/training-api.md |
| Training API losses | ../fireworks-training/references/training-api-losses.md |
| Secure training | ../fireworks-training/references/secure-training-operations.md |
| Models and shapes | ../fireworks-training/references/models-shapes-and-cost.md |
| Run state and resume | ../fireworks-training/references/run-state-and-reporting.md |
| Deployment and teardown | ../fireworks-training/references/deploy-and-troubleshoot.md |
| Error catalog | ../fireworks-training/references/error-reference.md |
| SDK and cookbook recipes | ../fireworks-training/references/sdk-recipes.md |
| SDK setup and examples | ../fireworks-training/references/sdk-setup.md, ../fireworks-training/references/sdk-examples.md |
| SDK migration and shapes | ../fireworks-training/references/sdk-migrate.md, ../fireworks-training/references/sdk-shapes.md |
| Checkpoints and tools | ../fireworks-training/references/sdk-checkpoints.md, ../fireworks-training/references/sdk-tools.md |
| Distillation | ../fireworks-training/references/sdk-distillation.md |
| RL losses | ../fireworks-training/references/rl-loss-paths.md, ../fireworks-training/references/rl-custom-loss.md, ../fireworks-training/references/rl-gradient-accumulation.md |
| Async and agentic RL | ../fireworks-training/references/rl-async.md, ../fireworks-training/references/rl-agentic.md, ../fireworks-training/references/rl-concurrency.md, ../fireworks-training/references/rl-dynamic-filter.md |
| Async RL metrics | ../fireworks-training/references/async-rl-metrics.md |
| Client phase tracing and Perfetto export | ../fireworks-training/references/client-phase-tracing.md |
| Hotload and sampling | ../fireworks-training/references/rl-hotload.md, ../fireworks-training/references/rl-sampling-timeouts.md |
| Renderer work | ../fireworks-training/references/renderer.md |
| Renderer verification | ../fireworks-training/references/renderer-verification.md |
| Journey telemetry | ../fireworks-training/references/telemetry.md |
| Telemetry notice | ../fireworks-training/references/telemetry-notice.md |