Common workflows
Workflow: mount a session and work locally
- Obtain the session ID.
- Set
AFS_AETHER_SERVER_ENDPOINTor[bridge].server_endpoint. - Mount the session locally.
- Use your editor, shell, or tooling against the mounted path.
- Use API or product workflows for checkpoints, approvals, or persistence as needed.
Workflow: switch between environments
When switching between staging, beta, or production-style environments:
- Keep endpoint selection explicit.
- Use shell wrappers or per-environment config files.
- Avoid reusing ambiguous cache locations unless that is intentional.
A safe shell pattern is:
export AFS_AETHER_SERVER_ENDPOINT='https://grpc.aetherfs.io'
export AFS_AETHER_CACHE_DIR="$HOME/.cache/aether-prod"
Workflow: enable debug output for an issue
When a mount or connectivity problem occurs:
export AFS_AETHER_LOG='debug'
export AFS_AETHER_LOG_FORMAT='text'
aether doctor
If another system will parse the output, prefer JSON logs instead.
Then rerun the actual mount command with explicit --session-id and --mount-dir, because doctor only covers local preflight checks.
Workflow: inspect runtime state after starting Aether CLI
Use commands such as:
aether status --session-id '<session-id>'
aether metrics show
aether logs tail
Remember that aether status is most useful when the local Aether CLI runtime is already active.
Workflow: isolate automation from your personal cache
For scripts or CI:
export AFS_AETHER_CACHE_DIR="$HOME/.cache/aether-ci"
export AFS_AETHER_CACHE_SHARED_ROOT="$HOME/.cache/aether-ci-shared"
That keeps automation state separate from your day-to-day workstation cache.
Workflow: tune metadata freshness
If your workflow needs fresh remote visibility, reduce TTL-related settings:
AFS_AETHER_LOOKUP_TTL_SECSAFS_AETHER_NEGATIVE_LOOKUP_TTL_SECSAFS_AETHER_DIR_CACHE_TTL_SECS
If your workflow prioritizes fewer round trips, increase them carefully.
Workflow: hand off to API-driven persistence
A common pattern is:
- Use Aether CLI for local editing.
- Use the HTTP API or your product workflow for annotations, approvals, checkpoints, and commits.
- Stop the local runtime when the interactive phase is complete.
This keeps local filesystem work and durable workflow control cleanly separated.