burn
Skill9 starsChanged 2 months agoMITView on GitHub
Use idle Claude Code subscription quota on real work up to a threshold, or monitor the 5-hour window. Use when the user says "burn quota", "use my remaining session", "prepare my reviews with burn", "watch my usage", or asks to set up/run the dont-burn-it-all tool in this repo.
- Reads credentials
2 other projects use a version of this.
Adopted by
Most recently taken up by
- yiyaw-lab/agent-armortheir own version, first seen Jun 22, 2026
- Fisher521/burn-claude-skilltheir own version, first seen Apr 16, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- yiyaw-lab/agent-armor★ 1same name
- Fisher521/burn-claude-skill★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repositoryOne other file it gives its agents.
- burn-jiraSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 17
If it prints MISSING, ask the user how to install it (do not install silently) and offer:
- Very specific: line 29
STOP — do not try to burn.
- Very specific: line 60
With Monitor: run burn as a background job (run_in_background: true), then use the
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
burn is a CLI that spends your subscription 5-hour quota on real headless work and stops at a
threshold. Drive it in three steps; never re-implement its logic in the loop.
1. Ensure burn is installed, then configured
Locate the binary, preferring an installed one:
command -v burn || { test -x ./burn && echo "./burn"; } || echo "MISSING"If it prints MISSING, ask the user how to install it (do not install silently) and offer:
- mise (prebuilt, no Go):
mise use -g "github:tbobm/dont-burn-it-all[exe=burn]@latest" - from source (in this repo):
go build -o ./burn .— then use./burn
Once you have a working burn command, verify config and relay the checklist:
burn setupOn any [FAIL] (no claude, no OAuth token, endpoint unreachable), report the remediation and
STOP — do not try to burn. [warn] lines are fine to proceed past.
2. Pick a mode
- Process a backlog (one session per Jira ticket, unattended): this is
burn process jiraand it has its own skill — useburn-jirainstead of this one. Anything about working labelled tickets, running while OOO/overnight/at the weekend, or reporting on such a run belongs there. - Launch (spend quota on a task): needs
--goal. Each launch runs--jobssessions to completion and stops — it does not loop. Start each launch deliberately.
Sessions prompt for permissions unless you passburn --goal "<real task>" --jobs 4 --target 80--dangerously-skip-permissions(opt-in; only with a throwaway--workdir). - Watch (governor only): polls usage every ~180s and notifies at the target, spawning nothing.
burn --watch --target 80
The first launch runs a ~3-minute preflight proving sessions bill to the subscription (not API).
This is intentional. If it hard-aborts, relay the fix (export CLAUDE_CODE_OAUTH_TOKEN=$(claude
setup-token), unset ANTHROPIC_API_KEY) and stop.
3. Run it without blocking the loop (sub-agent + Monitor)
burn --watch and multi-job launches are long-running. Do NOT sit in a foreground wait.
- As a sub-agent: the main loop can dispatch this skill to a sub-agent (Agent tool) so
burnruns isolated while the main session continues. The sub-agent runs the steps above and returns a summary (final 5-hour %, sessions run, errors). - With Monitor: run
burnas a background job (run_in_background: true), then use the Monitor tool with an until-loop to wait cheaply for a condition instead of polling in context — e.g. until the background job exits, or until the last line of~/.claude/burn/worker.jsonlshowsfive_hour_before >= target. Report when the condition trips.
Pick a --target above the current 5-hour % (see burn --dry-run --goal x); a target below it
makes launch refuse immediately by design.
4. Reporting
burn overview summarizes past sessions grouped by goal; burn overview --group item breaks a
burn process run down per work item. BURN_NOTIFY_CMD forwards a target-hit alert anywhere
($BURN_MSG via sh -c).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

