agentleFS
junhjangjunhjang/claude-harness/.claude/skills/budget-status/SKILL.md

budget-status

Skill0 starsChanged 6 months agoMITView on GitHub

Report daily LLM token budget — today's consumption per tier (Haiku/Sonnet/Opus), the configured cap, percentage used, time until UTC reset, per-source breakdown, and whether the cron kill flag is set. Run when asked about LLM cost / budget / token consumption / how much spent today, when about to fire an LLM-bearing slash, or when the operator asks if it is safe to run another costly job. Read-only ($0, no LLM calls), reads var/budget/today.json + var/cron.killed flag.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check whether the budget store exists:

    line 30
  2. Very specific:

    If consumed ≥ 80% of cap → warn user.

    line 50
  3. Very specific:

    If consumed ≥ 100% of cap → state that cron jobs should self-disable per the cost-discipline principle in docs/harness/principles.md.

    line 51
  4. Distinctive:

    This skill never spends LLM tokens on its own: All reasoning here is reading files and arithmetic — no external calls.

    line 53

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Report harness LLM-cost status: today's token consumption per tier (Haiku / Sonnet / Opus), the daily cap, and whether the cron kill flag is set.

Storage contract: budget tracking lives at var/budget/today.json (gitignored). Schema (when wired):

{
  "date_utc": "2026-05-05",
  "cap_tokens": 5000000,
  "consumed": {
    "haiku":  { "input": 0, "output": 0 },
    "sonnet": { "input": 0, "output": 0 },
    "opus":   { "input": 0, "output": 0 }
  },
  "by_source": {
    "C1_log_scan": 0,
    "C2_docs_audit": 0,
    "user_fix_ticket": 0
  },
  "reset_at_utc": "2026-05-06T00:00:00Z"
}

When the user invokes /budget-status:

  1. Check whether the budget store exists:
    ls -la var/budget/today.json 2>/dev/null
  2. If absent: Report:
    • "Budget store not initialized at var/budget/today.json."
    • "No LLM-cost-incurring cron jobs are currently active."
    • "Manual /audit-* and /review-* invocations are not currently metered — usage cap relies on operator awareness."
    • "Schema target: see body of this skill for the planned format."
  3. If present: parse and report:
    • Today's consumed tokens per tier (Haiku/Sonnet/Opus, input+output split).
    • Daily cap and percentage used.
    • Time until reset (UTC midnight by default).
    • Per-source breakdown if available.
    • Whether a circuit-breaker file is present ([ -f var/cron.killed ]) — if so, prominently flag it.
  4. Threshold warnings (when the budget store exists):
    • If consumed ≥ 80% of cap → warn user; suggest the consuming repo's circuit-breaker slash (e.g., a /kill-cron-style command) if any cron is still running.
    • If consumed ≥ 100% of cap → state that cron jobs should self-disable per the cost-discipline principle in docs/harness/principles.md.

This skill never spends LLM tokens on its own. All reasoning here is reading files and arithmetic — no external calls.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.