Paperclip
Paperclip finds 822 million weekly tokens hiding behind a reported $0
A Paperclip patch says subscription-authenticated Claude runs emitted a numeric zero that the ledger accepted as authoritative cost. On the inspected instance, roughly 822 million tokens a week appeared free; the proposal separates actual cash, derived rate-card value and unpriced usage.
paperclipai/paperclip PR #10476 is the inspected primary source: “fix(costs): price subscription runs from a rate card instead of booking $0 as reported.” The two related records below were inspected as supporting context, not treated as independent confirmation.
The facts
- paperclipai/paperclip PR #10476 was created 2026-07-30T00:11:01Z and was open at inspection; its title is “fix(costs): price subscription runs from a rate card instead of booking $0 as reported.” - paperclipai/paperclip PR #10478 was created 2026-07-30T00:50:03Z and was open at inspection; its title is “feat: add authoritative issue PATCH receipts.” - paperclipai/paperclip PR #10472 was created 2026-07-29T23:25:15Z and was open at inspection; its title is “feat: skill usage analytics in the skill browser and Studio (experimental).” - The repository reported 75095 stars, 5007 open issues, default branch master, and last push 2026-07-30T01:01:17Z. - The latest tagged-release baseline checked was v2026.722.0 published 2026-07-22T23:05:41Z.
What changed
paperclipai/paperclip PR #10476 reports: The bug The Claude Code CLI emits a well-formed total_cost_usd: 0 when it runs under OAuth / subscription auth. The ledger's cost-status guard was a type check only: A numeric 0 is not null, so it sailed straight through and every subscription run was written as an authoritative cost_status='reported' zero. That is the worst of the three possible outcomes: not a null that reads as "we don't know", but a confirmed $0 — indistinguishable from genuinely free work. On the instance where this was found, that was ~822M tokens/week of
paperclipai/paperclip PR #10478 reports: Thinking Path - Paperclip is the open source app people use to manage AI agents for work. - Agents update tasks through the issue API. - The update response did not state which values changed. - Blocker updates also did not echo the scalar blocker IDs. - Agents therefore used an extra GET request to confirm a successful write. - This pull request adds an authoritative change receipt and an optional small response. - The benefit is fewer API calls with a clear and compatible write contract. Linked Issues or
paperclipai/paperclip PR #10472 reports: Thinking Path - Paperclip is the open source app people use to manage AI agents for work - Companies attach reusable skills to agents, browse them in the skill browser, and author them in the Skill Studio - Today a skill's "usage" only means config-based adoption — how many agents have it attached — plus social counters; there is no runtime signal showing which skills agents actually load or read during runs - Without that signal, operators cannot tell which skills earn their context cost, which are dead weight, and
Why it matters
A clean zero is more dangerous than missing data because dashboards and budget decisions treat it as settled fact. Subscription economics still need a notional usage value without pretending list price is cash spend. This is an operator boundary: persisted state, execution ownership and visible controls must describe the same event. A workaround that broadens access, deletes state or hides the symptom would trade a visible defect for an unmeasured one.
Current
The primary record was open when captured on 2026-07-30. The release baseline was v2026.722.0 published 2026-07-22T23:05:41Z. Current repository metadata, pull requests, issues, release records, Google News discovery, Hacker News discovery, Lobsters, Metamesh and the rendered ClawCharts row were checked. ClawCharts selected the subject; it did not prove the claim.
Evidence
The primary URL is https://github.com/paperclipai/paperclip/pull/10476. Supporting records are https://github.com/paperclipai/paperclip/pull/10478 and https://github.com/paperclipai/paperclip/pull/10472. Their source bodies, timestamps and states were captured in the daily evidence bundle. Test counts, reproductions and deployment observations remain attributed to their authors unless explicitly identified as independently rerun.
Source boundary
Open work is described as open, closed work as closed, and operator reports as reports. A pull request is evidence of proposed or reviewed direction, not proof of a shipped release. Issue closure would not by itself prove a deployed fix. Search residue, package mirrors and historically published source spines were excluded.
Operator take
Keep actual cost and rate-card value separate, backfill historical events from stored token counts, and make unknown models visibly unpriced rather than silently free. Preserve a before-state receipt, make the smallest reversible change, and verify the original failure independently after intervention.
Caveat
Public project records are mutable. Status, scope and evidence can change after publication. This brief records the inspected state and does not authorize changes to a reader’s deployment.
Source inspected; source state, environment and release boundary remain explicit.