BF ai-usage-ledger
Compile every locally recorded AI coding-agent model call (Claude Code, Codex CLI/Desktop, GitHub Copilot CLI, Gemini CLI, opencode, OpenClaw, Cline/Roo/Kilo, aider, Kimi Code, Mistral Vibe, Continue, pi, Codebuff and any JSON-logging tool) across the operator's own profile and, when enabled, other drives, user profiles, WSL distros and named SSH hosts into one append-only, de-duplicated ledger stored as SQLite, JSON or CSV; price it at list API rates; split it by subscription account and billing plan; validate it against the tools' own counters; and render a branded light/dark dashboard plus a research-grade study package. Onboards once (branding, theme, storage backend, scheduled refresh, anonymised publication copy; hosts and accounts auto-detected per OS), remembers the preferences, refreshes with one command or on a schedule, renders fifteen ledger documents (statements, memos, briefs, billing evidence) as Markdown, HTML, PDF and DOCX, can anonymise everything for publication, optionally archives the raw log files themselves so they outlive the tools' retention, and answers detailed questions (per day, project, account, model or session, and inside prompt text or the archived logs when those opt-ins are on) through a query layer over the accumulated store. Disclosure: this skill reads agent transcripts and tool databases on this machine and on hosts the operator names, keeps a durable local ledger under ~/.ai-usage-ledger and the chosen working directory, and captures prompt text or credential-derived account claims only when the operator opts in. Use when asked to "find all my AI logs", "how many tokens did I use", "what would this have cost on the API", "compare my subscriptions", "usage by account", "cache savings", "set up my usage ledger", "monthly usage statement", "publish my usage anonymously", or to refresh an existing ledger.
Compile every locally recorded AI coding-agent model call (Claude Code, Codex CLI/Desktop, GitHub Copilot CLI, Gemini CLI, opencode, OpenClaw, Cline/Roo/Kilo…
As a process F 31/100 · Will not run — References files that are not bundled: assets/examples/example-onboarding.md, assets/logo.png
How to improve
- Shorten the description to 1024 characters.
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- The text references files that are not there: add them or drop the references.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 2
✓ No critical or high findings
Medium and low: 2
-
low Risky intent
intent-offensive-securityREADME.md:184Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| Not required | Outbound network access, credential use, persistence, privilege escalation, destructive file operations, background services. |
-
low Risky intent
intent-offensive-securitySKILL.md:146Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| Not required | Outbound network access (pricing pages are fetched by the operator, not the scripts), credential use, persistence, privilege escalation, destructive file operations, background servic
Files scanned: 42. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 1872 chars, limit 1024 - warning
body-longSKILL.md body ≈ 9281 tokens (recommended < 5000); move details to references/ - warning
missing-refreference to a missing file: assets/examples/example-onboarding.md - warning
missing-refreference to a missing file: assets/logo.png - note
description-budgetdescription takes 1872 of the ~15000-char shared budget for all skills
Process rating: all ten parameters 31/100
- 0Tools and files. 2 referenced file(s) missing: assets/examples/example-onboarding.md, assets/logo.png
- 0Steps. Prose only: no discrete steps
- 0Result and completion. Does not say what the result is
- 30Running it twice. 38 mutating operations with no state check
- 40Consistency. Frontmatter name (ai-usage-ledger) differs from the folder (ai-usage-ledger-skill)
- 40Execution cost. Instruction body is 9281 tokens: crowds the task out of the window
- 60Failures and branches. 2 branches
- 70When it triggers. States when to use, but not when not to
- 70Inputs and preconditions. Inputs and preconditions are listed
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 18 top-level sections: this looks like several domains in one skill
- low The response is described with custom markup (18 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1872: 120–800 characters recommended
- +3No numbered steps or checklist
- +3Output format is not stated: the model decides each time
- +4No input/output examples
- +2Single-language instructions
- +5Description quotes 9 example trigger phrases
- +4Structure: 19 headings
- +4Reference files are cited in the instructions (6 of 6)
- +3All 15 scripts are documented
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 40.