The core contribution is empirical, not the Rust crate: a catalog of 63 confirmed LLM-agent budget-overrun incidents across 21 orchestration frameworks (2023–2026), each backed by a quoted GitHub issue and (where reported) a dollar loss, classified at two-rater κ = 0.837 (N = 113). The crate is one case-study mitigation — affine ownership making clone/double-spend/use-after-delegation compile errors. Honest finding: on single-agent workloads a 4-line Python counter ties it 0/30; the affine type only pulls ahead under operator error in multi-agent delegation (the fan-out race is rejected by the borrow checker, while asyncio overshoots 30/30). Binary-level cap-soundness is left open. Full artifact (catalog CSV, crate, proofs, reproduce.sh): <a href=\"https://github.com/sajjadanwar0/token-budgets\" rel=\"nofollow\">https://github.com/sajjadanwar0/token-budgets</a><br>Feedback on the taxonomy welcome.</p>\n","updatedAt":"2026-06-04T16:51:53.296Z","author":{"_id":"648622c80e6657a2a9400677","avatarUrl":"/avatars/0b56f575f0556366b34f0157a6a613bf.svg","fullname":"Sajjad Khan","name":"sajjadanwar0","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":1,"identifiedLanguage":{"language":"en","probability":0.8308810591697693},"editors":["sajjadanwar0"],"editorAvatarUrls":["/avatars/0b56f575f0556366b34f0157a6a613bf.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.04056","authors":[{"_id":"6a2198e03490a593e87b0fab","name":"Sajjad Khan","hidden":false}],"publishedAt":"2026-06-02T00:00:00.000Z","submittedOnDailyAt":"2026-06-04T00:00:00.000Z","title":"Token Budgets: An Empirical Catalog of 63 LLM-Agent Budget-Overrun Incidents, with an Affine-Typed Rust Mitigation as a Case Study","submittedOnDailyBy":{"_id":"648622c80e6657a2a9400677","avatarUrl":"/avatars/0b56f575f0556366b34f0157a6a613bf.svg","isPro":false,"fullname":"Sajjad Khan","user":"sajjadanwar0","type":"user","name":"sajjadanwar0"},"summary":"LLM-agent budget overruns are a documented production failure class: a single retry loop can spend thousands of dollars before an operator notices, and the in-process integrity properties that would prevent it (no aliasing, no double-spend, no use-after-delegation of a cost-bearing value) are enforced, if at all, by ad-hoc wrappers rather than by the type system. Our central contribution is empirical: a catalog of 63 confirmed production incidents from 21 orchestration frameworks (2023-2026), each backed by a quoted GitHub issue and, where reported, a dollar loss, organized into an eight-cluster failure taxonomy (inter-rater Cohen's kappa = 0.837, N = 113), plus 47 supplementary structural entries. As one mitigation evaluated against this taxonomy, we build token-budgets, an 1,180-line Rust crate (no unsafe) that operationalizes affine ownership so that cloning, double-spending, or using a budget after delegating it are compile errors rather than runtime hazards an operator must remember to avoid. The dollar cap is runtime arithmetic under an estimator assumption; the affine layer makes that arithmetic non-bypassable. On single-agent workloads a 4-line Python counter matches the crate at 0/30 overshoot, so the distinguishing value is non-bypassability under operator error in multi-agent delegation: the delegation-fanout race documented in 11 incidents is rejected by the borrow checker at compile time, while the same pattern under asyncio overshoots 30/30 and three disciplined alternatives overshoot 0/30. Across five runtimes, three providers, and a temperature-stratified live-API test (N = 160), the approach reports zero cap violations and zero false refusals, at operational parity with concurrent work. Static over-reservation is 4-6x (2.11x adaptive). Binary-level cap-soundness on the running binary is left open.","upvotes":0,"discussionId":"6a2198e13490a593e87b0fac","githubRepo":"https://github.com/sajjadanwar0/token-budgets","githubRepoAddedBy":"user","githubStars":1},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.04056.md"}">
Token Budgets: An Empirical Catalog of 63 LLM-Agent Budget-Overrun Incidents, with an Affine-Typed Rust Mitigation as a Case Study
Abstract
LLM-agent budget overruns are a documented production failure class: a single retry loop can spend thousands of dollars before an operator notices, and the in-process integrity properties that would prevent it (no aliasing, no double-spend, no use-after-delegation of a cost-bearing value) are enforced, if at all, by ad-hoc wrappers rather than by the type system. Our central contribution is empirical: a catalog of 63 confirmed production incidents from 21 orchestration frameworks (2023-2026), each backed by a quoted GitHub issue and, where reported, a dollar loss, organized into an eight-cluster failure taxonomy (inter-rater Cohen's kappa = 0.837, N = 113), plus 47 supplementary structural entries. As one mitigation evaluated against this taxonomy, we build token-budgets, an 1,180-line Rust crate (no unsafe) that operationalizes affine ownership so that cloning, double-spending, or using a budget after delegating it are compile errors rather than runtime hazards an operator must remember to avoid. The dollar cap is runtime arithmetic under an estimator assumption; the affine layer makes that arithmetic non-bypassable. On single-agent workloads a 4-line Python counter matches the crate at 0/30 overshoot, so the distinguishing value is non-bypassability under operator error in multi-agent delegation: the delegation-fanout race documented in 11 incidents is rejected by the borrow checker at compile time, while the same pattern under asyncio overshoots 30/30 and three disciplined alternatives overshoot 0/30. Across five runtimes, three providers, and a temperature-stratified live-API test (N = 160), the approach reports zero cap violations and zero false refusals, at operational parity with concurrent work. Static over-reservation is 4-6x (2.11x adaptive). Binary-level cap-soundness on the running binary is left open.
Community
The core contribution is empirical, not the Rust crate: a catalog of 63 confirmed LLM-agent budget-overrun incidents across 21 orchestration frameworks (2023–2026), each backed by a quoted GitHub issue and (where reported) a dollar loss, classified at two-rater κ = 0.837 (N = 113). The crate is one case-study mitigation — affine ownership making clone/double-spend/use-after-delegation compile errors. Honest finding: on single-agent workloads a 4-line Python counter ties it 0/30; the affine type only pulls ahead under operator error in multi-agent delegation (the fan-out race is rejected by the borrow checker, while asyncio overshoots 30/30). Binary-level cap-soundness is left open. Full artifact (catalog CSV, crate, proofs, reproduce.sh): https://github.com/sajjadanwar0/token-budgets
Feedback on the taxonomy welcome.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.04056 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.04056 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.04056 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.