CLI reference
Every sparetokens command and flag, with examples.
The worker installs one command, sparetokens. This reference lists every command and flag, taken from the worker's own command tree. Run sparetokens --help or sparetokens COMMAND --help for the same text on your machine.
The worker is not public yet, so these are beta steps. Install today is from a checkout (pip install . or pipx install .). Available at launch Install with pipx install sparetokens, Homebrew or the one-line installer; see install.
Commands at a glance
sparetokens register- create your public handle (username) and a contributor token
sparetokens link- add this machine to your account with a code from your dashboard on sparetokens.ai
sparetokens web-link- get a one-time code that joins your website sign-in to this computer's account
sparetokens login- sign in with GitHub (device code; no password goes through SpareTokens)
sparetokens run- claim tasks and work on them
sparetokens status- what each source will do next and why, plus registration and which providers are ready
sparetokens give- choose how a source gives: --fixed, --expiring, --share, --all, --idle (or --leftovers where a provider allows it); writes rules.toml
sparetokens boost- a one-off boost: give an amount now (boost openai 5) or run for a while (boost mac 2h); it expires
sparetokens preview- what a source's rule would have done over a week: given, expired and kept (an example week, clearly labelled)
sparetokens config- show or edit local config (references to credentials, never the credentials)
sparetokens profile- choose how you appear on the public leaderboard and which fields are public
sparetokens setup- guided setup in nine steps (safe defaults; Enter keeps them)
sparetokens rotate-token- replace your worker token (or, with --account, your account token) now; `run` also does this on its own every 30 days
sparetokens pause- pause donating (all sources), optionally for a while: pause 2h
sparetokens remote- phone app control: on, off, status, or allow a bounded remote boost
sparetokens resume- resume donating after pause
sparetokens logs- show the background service log
sparetokens service- install, remove or check the background service
sparetokens uninstall- remove the service, config and local data (asks what to do with your public profile)
sparetokens detect- list the logins, keys and local model servers found on this machine
sparetokens doctor- self-check: OS, sandbox, providers found, provider-terms status, config dir and whether the coordinator answers (read-only)
sparetokens statusline- opt-in Claude Code status line hook that records your plan usage locally
sparetokens inspect- a task in plain language before it runs: goal, exact prompt, tools, network, budget, checks and signatures
sparetokens verify-log- download the public task log and check its hash chain, Merkle root and signed head
sparetokens report- report a task that looks wrong, harmful or wasteful (goes to the public report queue, redacted and shown after review)
sparetokens review- review mode: be asked before each new task type runs for the first time
sparetokens history- recent donations from the local ledger; --show TASK prints that task's local transcript
sparetokens budget- what has been spent today, this week and this month, per source
sparetokens agent-token- let your own AI agent contribute: create, list, pause or revoke its scoped, expiring token (needs your account token)
sparetokens mcp- run the local MCP setup assistant (stdio only) so your own AI agent can set SpareTokens up, check status and adjust giving rules
sparetokens feedback- send a note to the people running the beta, or open a pre-filled GitHub issue (bug, idea or goal proposal)
Global option: --version prints the installed version.
Command reference
sparetokens register
Create your public handle (username) and a contributor token.
--handle- your public username: 3-24 letters, digits, _ or -; shown as typed, unique ignoring case Required.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost. Default:
http://127.0.0.1:8795. --public- opt in to appearing by handle on the public leaderboard (default: anonymous)
--replace- this computer already has a worker: register a new one anyway (the old one stops being used here)
The first registration creates an account and a worker. Running it again on a registered machine changes nothing: it tells you the current handle and how to rename it with sparetokens profile --handle NAME. To start a separate second worker on the same machine, add --replace; the name is then only a private device label and your public handle stays the same. A taken handle returns an error with a free suggestion.
sparetokens register --handle yourname --public
sparetokens link
Add this machine to your account with a code from your dashboard on sparetokens.ai.
code- the short code shown on your dashboard, for example ABCD-2345 (valid for 10 minutes, single use) Required.
--label- private name for this machine (default: from the hostname)
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
--replace- replace a worker already registered on this machine
sparetokens link ABCD-2345
sparetokens web-link
Get a one-time code that joins your website sign-in to this computer's account.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
sparetokens login
Sign in with GitHub (device code; no password goes through SpareTokens).
--label- private name for this machine (default: from the hostname)
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
sparetokens login
sparetokens run
Claim tasks and work on them.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
--provider {anthropic-api|claude-code|openai-api|codex|xai|openrouter|local|none}- where the model comes from; every provider keeps its credentials on this machine. 'none' uses no model and only runs solver tasks Default:
anthropic-api. --budget-usd- stop once this much has been spent (providers that report dollars; also caps each task)
--token-budget- stop after about this many tokens (works for every provider that reports tokens)
--time-budget-min- stop after this many minutes of model work (useful for local models)
--tier {1|2|3}- capability tier you declare for this model (1 small/local, 2 mid, 3 frontier); default depends on provider and model
--base-url- OpenAI-compatible endpoint for --provider local (default http://127.0.0.1:11434/v1, Ollama)
--price-in-per-mtok- openai-api or xai: your input price per million tokens, to enable dollar accounting
--price-out-per-mtok- openai-api or xai: your output price per million tokens
--source- name of the configured source this run belongs to, for the local ledger (default: the provider name)
--now- a source with an idle rule: start now instead of waiting for the machine to be idle (power and temperature checks still apply)
--scheduled- evaluate the source's rule as the background service would (a manual rule then refuses)
--max-tasks- Stop after this many tasks. Default:
1. --model- model id for the chosen provider, e.g. claude-sonnet-5-5, gpt-..., or a local model name
--types- only these task types
--allow-bash- let the agent run sandboxed shell commands (needed for compute tasks)
--no-web- do not allow web search/fetch (skips literature tasks)
--task-cap-multiplier- credit-burn mode: multiply each task's suggested budget cap (still bounded by --budget-usd) Default:
1.0. --model-hints- use a task's suggested model (e.g. Opus for referee tasks) instead of --model; off by default
--no-model-hints- always use --model (the default; kept for older scripts)
--json-summary- write total_cost_usd, tasks attempted/submitted/verified and per-task details to this path
--workdir- Parent directory for the temporary task directories. Default:
~/.sparetokens/work. --keep-workdir- Keep each task directory after the run, for inspection.
--wait-verify- poll up to N*2s for the verification outcome (0 to skip) Default:
30.
Without a configured source only the flags you pass apply. With --source NAME the source's rule and caps apply too. A manual run does not check the global pause. --provider none needs a working OS sandbox and runs only solver tasks.
sparetokens run --budget-usd 2 --max-tasks 3 --allow-bash
sparetokens run --provider none --allow-bash --max-tasks 5
sparetokens run --provider local --model qwen3:30b --types compute --max-tasks 1
sparetokens status
What each source will do next and why, plus registration and which providers are ready.
--json- the same as JSON, for agents and scripts
For each source one plain sentence says what it will do next and why, in the order work is offered. --json gives the same fields for agents: state, will_run, sentence, reason, layer, next_change_in_s, headroom_usd and boost.
sparetokens status
sparetokens status --json
sparetokens give
Choose how a source gives: --fixed, --expiring, --share, --all, --idle (or --leftovers where a provider allows it); writes rules.toml.
source- a source name, or a short name like openai, anthropic, xai, openrouter or mac (a new source is created) Required.
--fixed- a fixed amount per period, a hard ceiling: 5 (dollars) or 10%% of --of
--expiring- use up expiring credit: the balance in dollars, with --by DATE; paced to the date, stops at --floor
--share- a steady percentage of a budget per period, spread evenly: 10%% with --of 100
--all- everything spare; API keys need --ceiling, and the keep-back cannot go below 5%%
--idle- idle time: only when the machine is idle, plugged in and cool (local models)
--leftovers- only what would expire before an allowance resets (only for providers whose terms allow it)
--by- expiring credit: the day it ends, like 2026-11-30
--floor- expiring credit or everything spare: always keep this many dollars
--of- the budget that a --share or a %% amount applies to
--period {day|week|month}- period for --fixed and --share (default month)
--window- leftovers: start giving this long before reset, like 6h
--keep- keep back this share of an allowance for yourself, like 15%%
--ceiling- a hard dollar ceiling on top of any style (required for --all on an API key)
--ceiling-period {day|week|month}- period for --ceiling (default month)
--nights- only 22:00 to 07:00
--hours- only between these local times
--quiet-hours- never between these local times
--idle-minutes- idle: minutes without keyboard or mouse first (default 10)
--any-power- idle: also run on battery
--key-ref- where the key lives: env:NAME or keychain:ITEM (a name, never the key)
Writes rules.toml in the config directory (see giving styles), after checking it. Exactly one style flag is required. --leftovers is refused for a provider whose terms do not allow automated use of a subscription login. The key is only ever referenced by name with --key-ref.
sparetokens give anthropic --fixed 5 --key-ref env:ANTHROPIC_API_KEY
sparetokens give promo --provider anthropic-api --expiring 30 --by 2026-11-30 --floor 2
sparetokens give xai --all --ceiling 20 --nights
sparetokens give mac --idle --nights
sparetokens boost
A one-off boost: give an amount now (boost openai 5) or run for a while (boost mac 2h); it expires.
source- Required.
what- dollars like 5 (valid 24 hours) or a duration like 2h
--cancel- end a boost early
A plain amount is dollars and lasts 24 hours; a duration such as 2h or 30m has no dollar limit. A boost replaces the schedule and the style, never the protections, keep-back or ceilings.
sparetokens boost anthropic 5
sparetokens boost mac 2h
sparetokens boost mac --cancel
sparetokens preview
What a source's rule would have done over a week: given, expired and kept (an example week, clearly labelled).
source- Required.
--json- Print the result as JSON.
Simulates the last 7 days and prints given, expired and kept. Your own usage is not recorded, so it runs on a clearly labelled example week, and also shows what your ledger recorded as given.
sparetokens preview promo
sparetokens preview mac --json
sparetokens config
Show or edit local config (references to credentials, never the credentials).
sparetokens config show
sparetokens config show
.
sparetokens config show
sparetokens config path
.
sparetokens config path
sparetokens config add-source
Add a source; --credential is env:NAME or keychain:ITEM.
name- The name of the source. Required.
--provider {anthropic-api|claude-code|openai-api|codex|xai|openrouter|local}- Which provider to use. Required.
--credential- env:VAR_NAME or keychain:ITEM_NAME (the secret itself is never stored)
--model- Model id for this source or run.
sparetokens config add-source work-api --provider anthropic-api --credential env:ANTHROPIC_API_KEY
sparetokens config set-rule
Set a source's donation rule and hard caps.
name- The name of the source. Required.
--type {manual|fixed_budget|percentage|idle_only|use_it_or_lose_it}- Rule type: manual, fixed_budget, percentage, idle_only or use_it_or_lose_it. Prefer `sparetokens give` for the newer styles. Required.
--period {day|week|month}- Period for fixed_budget and percentage rules.
--amount-usd- Dollars per period for a fixed_budget rule.
--percent- Percentage of the source's budget per period, for a percentage rule.
--of-usd- The dollar budget that --percent applies to.
--idle-minutes- Idle minutes required before an idle_only rule may run.
--hours-before-reset- For use_it_or_lose_it: only donate in the last N hours before the allowance resets.
--reserve-percent- For use_it_or_lose_it: the share of the allowance to keep for yourself.
--per-task-usd- Hard cap in dollars for one task.
--per-day-usd- Hard cap in dollars per day for this source.
--amount-tokens- Tokens per period for a fixed_budget rule.
--of-tokens- The token budget that --percent applies to.
--per-task-tokens- Hard cap in tokens for one task.
--per-day-tokens- Hard cap in tokens per day for this source.
--max-turns- Maximum agent turns per task.
--timeout-s- Timeout in seconds per task.
--window- use_it_or_lose_it window: five_hour or seven_day (Claude), primary or secondary (Codex)
sparetokens config set-rule work-api --type fixed_budget --amount-usd 5 --period month
sparetokens config set-rule laptop --type idle_only --idle-minutes 10
sparetokens config set-models
Choose models for a source: best, cheapest that qualifies, or a list.
name- The name of the source. Required.
--mode {best|cheapest|list}- Required.
--model- repeat for each model (required for mode list)
sparetokens config set-models work-api --mode cheapest
sparetokens config models
Show the model a source would use and which task types each model unlocks.
name- The name of the source. Required.
sparetokens config models work-api
sparetokens config explain
What the rules would do right now.
name- The name of the source.
sparetokens config explain
sparetokens config reset-breaker
Resume a source that the circuit breaker or spend reconciliation paused.
name- The name of the source. Required.
sparetokens config reset-breaker work-api
sparetokens config remove-source
.
name- The name of the source. Required.
sparetokens config remove-source work-api
sparetokens profile
Choose how you appear on the public leaderboard and which fields are public.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
--display-mode {anonymous|handle|first_last_initial|full_name}- How you appear: anonymous, handle, first_last_initial or full_name.
--handle- your public username (3-24 letters, digits, _ or -); also your profile URL name in every non-anonymous mode
--first-name- real name, shown only if you choose first_last_initial or full_name
--last-name- Real name, shown only if you choose first_last_initial or full_name.
--link- https URL, shown only with --show-link
--country- two-letter code, shown only with --show-country
--hardware-label- a short label such as 'Mac mini' sent with results; public only with --show-hardware ('' clears it)
--show-link, --no-show-link- make link public (or --no-show-link)
--show-country, --no-show-country- make country public (or --no-show-country)
--show-models, --no-show-models- make models public (or --no-show-models)
--show-providers, --no-show-providers- make providers public (or --no-show-providers)
--show-hardware, --no-show-hardware- make hardware public (or --no-show-hardware)
--show-tokens, --no-show-tokens- make tokens public (or --no-show-tokens)
sparetokens profile --display-mode first_last_initial --first-name Ada --last-name Lovelace
sparetokens profile --show-models
sparetokens setup
Guided setup in nine steps (safe defaults; Enter keeps them).
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
--yes- accept every default without asking
--skip-test- skip the live test task
Nine steps: identity, how you appear, what to connect, how much, which models, what to work on, when, a live test, and a summary. Every question has a safe default. --yes with nothing detected connects nothing and spends nothing.
sparetokens setup
sparetokens setup --yes --skip-test
sparetokens rotate-token
Replace your worker token (or, with --account, your account token) now; `run` also does this on its own every 30 days.
--account- rotate the account token in account.json instead (copy the new file to your other machines)
--resume-only- with --account: only finish an interrupted rotation
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
Replaces your worker token with a new one. run does this by itself when the token is older than 30 days (set SPARE_TOKEN_ROTATE_DAYS, or 0 to turn it off). The new token is saved to worker.json before the old one is dropped, so an interruption cannot lock the worker out: the next run finishes the job. A token that comes from SPARE_TOKEN is never rotated, because it cannot be saved back. --account rotates the account token in account.json the same way; copy the new file to any other machine that uses it, because the old account token stops working at once. If a rotation fails, the command says the token was not changed and this machine keeps the one it had; if the reason is a 401, run sparetokens setup or sparetokens link CODE --replace.
sparetokens rotate-token
sparetokens rotate-token --account
sparetokens pause
Pause donating (all sources), optionally for a while: pause 2h.
duration- 30m, 2h, 1d; omit to pause until resumed
sparetokens pause 2h
sparetokens remote
Phone app control: on, off, status, or allow a bounded remote boost.
action {status|on|off|boost|boost-off}- install, uninstall or status. Default: status. Default:
status. source- for boost: the source a remote boost may use
--max-hours- for boost: the longest a remote boost may run (up to 8)
--max-usd- for boost: the most a remote boost may give, in dollars (up to 50)
sparetokens resume
Resume donating after pause.
--breaker- instead, clear the circuit-breaker pause of one source (same as `config reset-breaker SOURCE`)
sparetokens resume
sparetokens logs
Show the background service log.
-n- Number of lines to show. Default 50. Default:
50. -f, --follow- Keep following the log as it grows.
sparetokens logs -n 100 -f
sparetokens service
Install, remove or check the background service.
action {install|uninstall|status}- install, uninstall or status. Default: status. Default:
status.
install writes a launchd agent on macOS or a systemd user unit on Linux. Nothing starts unless you ask for it.
sparetokens service install
sparetokens service status
sparetokens uninstall
Remove the service, config and local data (asks what to do with your public profile).
--profile {delete_profile|keep_anonymous}- delete_profile: delete your public profile (results stay as 'Former contributor'); keep_anonymous: keep your verified results (the default with --yes)
--yes- skip the confirmation prompts (keeps results anonymously unless --profile is given)
Asks you to type uninstall unless you pass --yes. With --yes and no --profile, your results are kept anonymously and the command says so. Without a terminal and without --yes it stops before removing anything. At the end it points to the dashboard and delete my data for exporting or deleting everything else.
sparetokens uninstall
sparetokens detect
List the logins, keys and local model servers found on this machine.
--json- Print the result as JSON.
sparetokens detect --json
sparetokens doctor
Self-check: OS, sandbox, providers found, provider-terms status, config dir and whether the coordinator answers (read-only).
--json- Print the result as JSON.
--offline- do not contact the coordinator
A read-only self-check for installs and bug reports: your OS, Python, whether a sandbox is available, which providers it found, the provider-terms status, the config directory and whether the coordinator answers. On macOS it suggests a keychain: reference when a source reads its key from an env: variable. It never prints a credential and changes nothing. --offline skips the network check.
sparetokens doctor
sparetokens doctor --offline --json
sparetokens statusline
Opt-in Claude Code status line hook that records your plan usage locally.
--print-settings- print the settings.json block that enables it
--chain- your existing status line command; its output is shown and it receives the same input
sparetokens statusline --print-settings
sparetokens inspect
A task in plain language before it runs: goal, exact prompt, tools, network, budget, checks and signatures.
task- Required.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
Shows the task in plain language before it runs: goal, the exact prompt, tools, network hosts, budget cap, how the result is checked, and the signatures and public log entry it checked. It asks the coordinator you are registered with. If a signature, template or log check fails it prints NOT VERIFIED, exits with status 3 and suggests a report. The worker makes the same checks before every task. See trust and verification.
sparetokens inspect 1234
sparetokens verify-log
Download the public task log and check its hash chain, Merkle root and signed head.
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
--mirror- check the public mirror instead, e.g. https://sparetokens.ai/data/tasklog
Downloads the public task log and checks the hash chain, the Merkle root and the signed head, and prints a summary or FAILED (exit status 3). With --mirror it reads head.json and entries/N.json from the public mirror instead of the coordinator.
sparetokens verify-log
sparetokens verify-log --mirror https://sparetokens.ai/data/tasklog
sparetokens report
Report a task that looks wrong, harmful or wasteful (goes to the public report queue, redacted).
task- Required.
--reason {harmful|not-research|privacy|wasteful|broken|other}- Required.
--text- what you saw (do not include keys or personal data; links and addresses are removed from the public copy)
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
Needs a registered worker. The report goes to the public report queue. The text you write is shown there with links and email addresses removed, once we have reviewed and approved it. Do not include keys or personal data.
sparetokens report 1234 --reason wasteful --text "The examples look odd"
sparetokens review
Review mode: be asked before each new task type runs for the first time.
--on--off--allow- always run this task type
--deny- never run this task type
--forget- ask again next time
Review mode asks you before each new task type runs for the first time, then remembers your answer in review.json in the config directory. With no flags it prints the current mode and your saved answers. When nobody is at the keyboard (a scheduled run), a new task type waits instead of running.
sparetokens review --on
sparetokens review --allow code_fill.abundant_count
sparetokens review --deny literature_status
sparetokens review --forget literature_status
sparetokens review --off
sparetokens history
Recent donations from the local ledger; --show TASK prints that task's local transcript.
--limit- How many entries to show. Default 20. Default:
20. --source- Limit to this source.
--show- print everything sent to and received from the AI for this task (kept 30 days, never uploaded)
With --show TASK it prints everything the worker sent to the AI and received back for that task. Transcripts stay on your machine for 30 days and are never uploaded.
sparetokens history --limit 10
sparetokens history --show 1234
sparetokens budget
What has been spent today, this week and this month, per source.
--source- Limit to this source.
sparetokens budget
sparetokens agent-token
Let your own AI agent contribute: create, list, pause or revoke its scoped, expiring token (needs your account token).
action {create|list|pause|resume|revoke|save}- create a token, list yours, pause or resume one (pause also gives its open tasks back), revoke one for good, or save a token read from standard input Required.
id- the token id from `agent-token list` (pause, resume, revoke)
--label- a short name for the agent, like claude-desk (create, save)
--days- create: days until it expires (default 30, at most 365) Default:
30. --scopes- create: comma list of claim, submit, read (default all three)
--max-leases- create: tasks it may hold at once (1 to 3, default 2)
--rate- create: requests a minute it may make (1 to 120, default 30)
--save- create: write the token to agent.json (readable only by you) instead of printing it
--coordinator- Coordinator URL. Default is $SPARE_COORDINATOR, or http://127.0.0.1:8795. Must be https, except for localhost.
Agent tokens let your own AI agent claim and submit tasks under your account (see agent tokens). create, list, pause, resume and revoke need your account token (account.json); an agent token cannot manage tokens. A token is shown once, so use --save to write it to agent.json (readable only by you) instead of printing it. save reads a token you created on the dashboard from standard input, never from a command line argument. The MCP server reads SPARE_AGENT_TOKEN or agent.json. Pausing stops new work and gives open tasks back; revoking is final.
sparetokens agent-token create --label claude-desk --save
sparetokens agent-token list
sparetokens agent-token pause 1
sparetokens agent-token revoke 1
pbpaste | sparetokens agent-token save
sparetokens mcp
Run the local MCP setup assistant (stdio only) so your own AI agent can set SpareTokens up, check status and adjust giving rules.
--print-config {claude-code|codex|json}- print the install snippet for that client instead of starting the server
Starts a local MCP server on standard input and output only, with no network listener, so your own assistant can list sources, preview and set giving rules, pause and resume, and read the docs. It never accepts or returns API keys and every change needs confirm: true. It needs the MCP extra: pipx install '.[mcp]'. See For agents.
claude mcp add sparetokens -- sparetokens mcp
sparetokens mcp --print-config codex
sparetokens feedback
Send a note to the people running the beta. It goes through the coordinator you are registered with, is emailed to support, and a reply comes to your verified email if your account has one. Text that looks like a key or token is refused before anything is sent. With one of the words bug, idea or goal it instead opens a pre-filled GitHub issue with no secrets or logs.
message- your message, in quotes or as several words; or one of
bug,idea,goalfor a pre-filled GitHub issue (default:bug). --coordinator- coordinator URL (default: the one you registered with)
--title- A short title for the issue (GitHub issue form only).
--details- a short description to pre-fill (scanned for credentials; GitHub issue form only)
--print-only- print the issue link instead of opening a browser (GitHub issue form only)
sparetokens feedback "the wizard asked me the same question twice"
sparetokens feedback idea --print-only
Environment variables
SPARE_COORDINATOR- Default coordinator URL. Without it the worker uses
http://127.0.0.1:8795. Planned A public coordinator URL arrives with the beta. SPARE_TOKEN_ROTATE_DAYS- Age in days at which
runrotates the worker token. Default 30; 0 turns automatic rotation off. SPARE_ALLOW_INSECURE_COORDINATOR- Set to
1to allow a plainhttp://coordinator that is not on this machine, for a private network you control. Without it the worker refuses anything that is not https or localhost. SPARE_CONFIG_DIR- Where
worker.json,account.jsonand the config live. Default~/.config/sparetokens. SPARE_STATE_DIR- Where the local ledger, service log and notices live. Default
~/.local/state/sparetokens. ANTHROPIC_API_KEY,OPENAI_API_KEY,XAI_API_KEY,OPENROUTER_API_KEY- Your provider keys, read from the environment by the matching provider. They are never written to disk or sent to the coordinator.
Files
~/.config/sparetokens/worker.json: coordinator, handle and contributor token (mode 0600).~/.config/sparetokens/account.json: account token (mode 0600).~/.local/state/sparetokens/: the local ledger,service.logandnotices.log.~/.sparetokens/work: the default working directory for tasks.
Edit this page on GitHub: available when the repository opens.