cost-check
Skill0 starsChanged 4 months agoMITView on GitHub
Query the invocation/spend log for current API spend vs daily and monthly budget thresholds and report a projected month-end total
1 other project uses a version of this.
Adopted by
Most recently taken up by
- kitloomlk-stack/devops-cloud-kittheir own version, first seen Oct 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- kitloomlk-stack/devops-cloud-kit★ 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 repository29 other files it gives its agents.
- adopt-existing-projectSkill
- adrSkill
- branch-conflict-checkSkill
- ci-watchSkill
- compound-learningsSkill
- config-auditSkill
- coverage-ratchetSkill
- doc-syncSkill
- eval-harnessSkill
- flaky-triageSkill
- followup-trackingSkill
- incident-captureSkill
- kit-doctorSkill
- kit-updateSkill
- logging-initSkill
- migration-checkSkill
- new-project-bootstrapSkill
- nightly-auditSkill
- observability-checkSkill
- parallel-workSkill
- perf-budgetSkill
- pr-babysitterSkill
- pre-prSkill
- prompt-regressionSkill
- releaseSkill
- session-handoffSkill
- sync-healthSkill
- using-the-kitSkill
- writing-kit-skillsSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 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.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 23
Run this any time you want to know where the project stands against its API budget before triggering expensive work, or after a run that …
- Very specific: line 55
If the table turns out to be empty (every figure below is $0.0000), that's not necessarily healthy — verify the DB path is right and that …
- Very specific: line 136
Read the thresholds from the profile — budget.daily_usd and budget.monthly_usd in .claude/kit.yaml (defaults: 20 / 100).
- Distinctive: line 146
Do not run cost-heavy evals or multi-step agent chains if the verdict is WARN or PAUSE — escalate to the user first.
- Very specific: line 148
If verdict is PAUSE, BudgetGuard.check() is already raising BudgetPausedError on every outbound call.
≈ “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
Cost Check
Applies when capabilities.llm.enabled is true (else skip, report N/A).
Profile-driven. The spend table is capabilities.llm.spend_table from .claude/kit.yaml (default: invocation_log). Substitute that name in every query below if your project uses a different value.
Price sanity check (update as pricing changes). claude-opus-5 $5/$25 per
MTok · claude-sonnet-5 $3/$15 (intro $2/$10 through 2026-08-31) ·
claude-haiku-4-5 $1/$5 · claude-fable-5 $10/$50. Use these only to
sanity-check the cost_usd figures below — the spend table's cost_usd
column stays authoritative.
Run this any time you want to know where the project stands against its API
budget before triggering expensive work, or after a run that felt heavy. This
skill only reads and reports — it never writes to the spend table or adjusts
budget thresholds itself. It reads directly from the invocation_log table —
the same source BudgetGuard uses at call time — so the numbers are
authoritative. For automated enforcement (raising before a call goes out), see
templates/cost_guard.py.
The queries below assume SQLite. Adjust
date()andstrftime()calls for Postgres (current_date,date_trunc,extract). SetDBto your database path or connection string.
Workflow
1. Set the database path
# kit.yaml → budget.spend_db_path (empty → discover the DB at the repo root)
DB="$(python3 - <<'PY'
import glob, os
# Prefer the profile value; else find a *.db at the repo root.
val = "" # substitute budget.spend_db_path from kit.yaml
if val:
print(val)
else:
cands = glob.glob("*.db")
print(cands[0] if cands else "")
PY
)"
[ -z "$DB" ] || [ ! -f "$DB" ] && echo "No spend DB found — set budget.spend_db_path in kit.yaml" >&2If the table turns out to be empty (every figure below is $0.0000), that's
not necessarily healthy — verify the DB path is right and that BudgetGuard /
your logging layer is actually writing rows, rather than assuming zero spend.
2. Today's total spend
# kit.yaml → capabilities.llm.spend_table (replace invocation_log below if different)
sqlite3 "$DB" "
SELECT printf('Today \$%.4f', COALESCE(SUM(cost_usd), 0.0))
FROM invocation_log
WHERE date(timestamp) = date('now');
"3. Today's per-task-type breakdown
sqlite3 -column -header "$DB" "
SELECT
task_type,
COUNT(*) AS calls,
SUM(tokens_in) AS tok_in,
SUM(tokens_out) AS tok_out,
printf('\$%.4f', SUM(cost_usd)) AS cost
FROM invocation_log
WHERE date(timestamp) = date('now')
GROUP BY task_type
ORDER BY SUM(cost_usd) DESC;
"4. Month-to-date total
sqlite3 "$DB" "
SELECT printf('Month-to-date \$%.4f', COALESCE(SUM(cost_usd), 0.0))
FROM invocation_log
WHERE timestamp >= date('now', 'start of month');
"5. Month-to-date per-model breakdown
sqlite3 -column -header "$DB" "
SELECT
model_alias,
COUNT(*) AS calls,
printf('\$%.4f', SUM(cost_usd)) AS cost
FROM invocation_log
WHERE timestamp >= date('now', 'start of month')
GROUP BY model_alias
ORDER BY SUM(cost_usd) DESC;
"6. 7-day rolling average and projected month-end
sqlite3 -column -header "$DB" "
SELECT
printf('\$%.4f', avg_daily) AS avg_daily_7d,
printf('\$%.4f', mtd + avg_daily * days_left) AS projected_month_end
FROM (
SELECT
(SELECT COALESCE(SUM(cost_usd), 0.0)
FROM invocation_log
WHERE timestamp >= date('now', 'start of month')) AS mtd,
(SELECT COALESCE(SUM(cost_usd), 0.0) / 7.0
FROM invocation_log
WHERE timestamp >= date('now', '-6 days')) AS avg_daily,
(CAST(strftime('%d',
date('now', 'start of month', '+1 month', '-1 day')) AS INTEGER)
- CAST(strftime('%d', date('now')) AS INTEGER)) AS days_left
);
"7. Read the thresholds and issue a verdict
Read the thresholds from the profile — budget.daily_usd and budget.monthly_usd
in .claude/kit.yaml (defaults: 20 / 100). Don't hunt for a project-specific
config file; the profile is the single source. Compare the numbers above:
- GREEN — today < 80% daily AND month-to-date < 80% monthly AND projection < monthly
- WARN — today ≥ 80% daily OR month-to-date ≥ 90% monthly OR projection ≥ monthly
- PAUSE — today ≥ daily limit OR month-to-date ≥ monthly limit (
BudgetGuardwill be raising)
Guardrails
- Do not run cost-heavy evals or multi-step agent chains if the verdict is WARN or PAUSE — escalate to the user first.
- If verdict is PAUSE,
BudgetGuard.check()is already raisingBudgetPausedErroron every outbound call; confirm before overriding the threshold.
Output
## Cost Check — <YYYY-MM-DD>
| Metric | Amount | Limit | % |
|--------------------------|-----------|-----------|--------|
| Today | $X.XXXX | $XX.XX | XX% |
| Month-to-date | $XX.XX | $XXX.XX | XX% |
| 7-day avg daily | $X.XXXX | — | — |
| Projected month-end | $XX.XX | $XXX.XX | XX% |
Top tasks today: <task_type ($X.XXXX), ...>
Top models MTD: <model_alias ($X.XXXX), ...>
Verdict: GREEN / WARN / PAUSE — <reason if not GREEN>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.

