disclose

Generates per-item AI usage disclosures from metered token and session ledger data.

1|Updated Aug 17, 2026
One-click install
npx skills add https://github.com/Jakeintech/waybill --skill disclose-jakeintech
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: disclose
Source: https://github.com/Jakeintech/waybill/tree/main/skills/disclose
Command: npx skills add https://github.com/Jakeintech/waybill --skill disclose-jakeintech

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Policies, clients, journals, and open-source projects increasingly ask "was AI used on this work, and how much?" Most people can only answer from memory. This Skill answers from the waybill ledger: per shipped item, it renders the recorded Claude role, metered sessions and tokens, token share, and evidence line — conservation-checked, never estimated. ## Core Features & Use Cases - Single-item disclosure: Produce a one-paragraph statement for a specific work item (e.g., "was AI used on PLAT-482") with recorded role, metered sessions/tokens, and shipping evidence. - Disclosure register: Build a window-wide table with one row per shipped item, including role, sessions, tokens, token share of the window, and an honest footer covering metering basis and unattributed spend. - Audience-aware rendering: Use --audience internal for org/client contexts with real identifiers, or --audience external for journals and OSS maintainers with pseudonymized identifiers. - Use Case: A client contract requires an AI involvement report for the quarter. Run the report query over the window and hand over a register where every number traces to metered transcripts, including items with no AI assistance recorded. ## Quick Start Ask the assistant to build your AI disclosure for a specific ticket or date range, for example: "was AI used on PLAT-482, and build my AI disclosure register for last quarter."

Frequently Asked Questions about disclose

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prove whether AI was used on a specific work item?▼

Query the ledger for the item's shipped row to get its recorded Claude role, metered sessions, and attributed tokens. The output is a one-paragraph disclosure citing the item key, PR, and metering basis, with event ids available on request.

How do I build an AI disclosure register for a client or policy?▼

Run the report query over a date window to produce one row per shipped item with role, sessions, tokens, and token share. Include items with no AI assistance so the register can prove a negative, and always show the unattributed spend percentage.

Can I share an AI disclosure externally without exposing ticket keys?▼

Yes, render the report with the external audience flag to pseudonymize identifiers while keeping numbers and roles intact. The named internal version remains available on request if the recipient needs it.

What if an item predates metering or its transcript was pruned?▼

The disclosure states that the meter cannot attest that item rather than guessing from memory. Zero metered tokens is a real, reportable answer; unmeasured periods are labeled as meter gaps.

Does token share measure how much of the code AI wrote?▼

No. Token share is the item's metered tokens divided by the window's metered spend, measuring assistance volume only. The Claude role is a recorded declaration, and the two are never blended into an invented AI authorship percentage.