Debug
Debug: triage a stuck or failed Fireworks training run.
Debug is read-only by default. Do not create jobs, upload data, or spend without
an explicit handoff to configure and a new approved plan.
Entry routing
| Signal |
Route |
| New training or approved retry |
configure |
| Unclear method, data, or starting example |
research |
| Failure, stall, bad quality, resume, or serving issue |
stay in debug |
Research owns the welcome-entry question and its privacy notice. Route a
vague first message there instead of asking a second welcome question.
Attribution and privacy
Reuse the run's FIREWORKS_SESSION_ID. If none exists, create one UUID. Set:
export FIREWORKS_CLIENT_SOURCE="fireworks-training-skill/2.2.0"
Record entry_skill: debug in the private run manifest. Before the first
structured question, show
../fireworks-training/references/telemetry-notice.md.
Check firectl skill-journey record --help once. When available, record only
registered IDs through ../fireworks-training/references/telemetry.md. If it is
unavailable or the user opts out, keep local run state and continue.
Never send raw errors, customer prose, credentials, datasets, or paths as
journey telemetry.
Triage
- Read
references/triage-paths.md.
- Ask one category question: job state, error, quality, resume/checkpoint, or
deploy/serving.
- Stop and wait for the answer.
- Follow the ordered read-only checks for that category.
- Use the detailed carrier references listed below.
Three-strike rule
After three failed hypotheses on the same issue:
- Stop guessing.
- Build an escalation bundle with UTC timestamps, resource IDs, model and
shape, CLI and SDK versions, cookbook commit, retry history, evidence, and
what was ruled out.
- Redact credentials and customer data.
- State what the evidence supports and what remains unknown.
Handoff
- Route to
configure only when the user wants a new or retry run.
- Route to
research when the starting example or method is wrong.
- On resolution without new spend, record
debug_resolved and stop.
Progressive references
| Need |
Reference |
| Triage categories |
references/triage-paths.md |
| Error catalog |
../fireworks-training/references/error-reference.md |
| Resume and idempotency |
../fireworks-training/references/run-state-and-reporting.md |
| Deployment proof |
../fireworks-training/references/deploy-and-troubleshoot.md |
| Journey telemetry |
../fireworks-training/references/telemetry.md |
Interaction rules
- Use AskQuestion one question at a time, then stop.
- Use plain customer language.
- The category question uses its five registered options.
- Do not retry or create resources from debug.
1---2name: debug3description: Diagnose Fireworks training and deployment issues, including stuck or failed jobs, error messages, poor quality, checkpoint and resume problems, and deployments that are ready but not serving. Use for systematic read-only triage before retrying. Use configure for a new or approved retry run; use research when the starting method, data, or cookbook example is unclear.4---56# Debug78`Debug`: triage a stuck or failed Fireworks training run.910Debug is read-only by default. Do not create jobs, upload data, or spend without11an explicit handoff to `configure` and a new approved plan.1213## Entry routing1415| Signal | Route |16|---|---|17| New training or approved retry | `configure` |18| Unclear method, data, or starting example | `research` |19| Failure, stall, bad quality, resume, or serving issue | stay in `debug` |2021Research owns the `welcome-entry` question and its privacy notice. Route a22vague first message there instead of asking a second welcome question.2324## Attribution and privacy2526Reuse the run's `FIREWORKS_SESSION_ID`. If none exists, create one UUID. Set:2728```bash29export FIREWORKS_CLIENT_SOURCE="fireworks-training-skill/2.2.0"30```3132Record `entry_skill: debug` in the private run manifest. Before the first33structured question, show34`../fireworks-training/references/telemetry-notice.md`.3536Check `firectl skill-journey record --help` once. When available, record only37registered IDs through `../fireworks-training/references/telemetry.md`. If it is38unavailable or the user opts out, keep local run state and continue.3940Never send raw errors, customer prose, credentials, datasets, or paths as41journey telemetry.4243## Triage44451. Read `references/triage-paths.md`.462. Ask one category question: job state, error, quality, resume/checkpoint, or47 deploy/serving.483. Stop and wait for the answer.494. Follow the ordered read-only checks for that category.505. Use the detailed carrier references listed below.5152## Three-strike rule5354After three failed hypotheses on the same issue:55561. Stop guessing.572. Build an escalation bundle with UTC timestamps, resource IDs, model and58 shape, CLI and SDK versions, cookbook commit, retry history, evidence, and59 what was ruled out.603. Redact credentials and customer data.614. State what the evidence supports and what remains unknown.6263## Handoff64651. Route to `configure` only when the user wants a new or retry run.662. Route to `research` when the starting example or method is wrong.673. On resolution without new spend, record `debug_resolved` and stop.6869## Progressive references7071| Need | Reference |72|---|---|73| Triage categories | `references/triage-paths.md` |74| Error catalog | `../fireworks-training/references/error-reference.md` |75| Resume and idempotency | `../fireworks-training/references/run-state-and-reporting.md` |76| Deployment proof | `../fireworks-training/references/deploy-and-troubleshoot.md` |77| Journey telemetry | `../fireworks-training/references/telemetry.md` |7879## Interaction rules80811. Use AskQuestion one question at a time, then stop.822. Use plain customer language.833. The category question uses its five registered options.844. Do not retry or create resources from debug.
fw-ai