> For the complete documentation index, see [llms.txt](https://docs.trueopen.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.trueopen.ai/protocol/v1/tasks/task-05-challenge.md). # Challenges Open, execute, and close an independent challenge verification round. The challenge window begins when round 1 closes. A challenge does not reverse a completed payment: no settlement has occurred yet. ```mermaid sequenceDiagram participant O as Challenge opener participant K as Chain participant B as Task Builders participant C as Eligible candidates participant V as Challenge Verifiers O->>K: Lock challenge bond and verification budget K->>K: Open next round and freeze prior-round facts K->>K: Beacon A builds challenge candidate window B-->>C: Broadcast challenge metadata C-->>B: Signed challenge handraise B->>K: Submit aggregated handraises opt Builders are unavailable C->>K: Candidate self-submits its handraise end K->>K: Exclude all previous-round participants K->>K: Beacon B selects challenge Verifiers V->>B: Fetch original Task data B-->>V: Input, output, and Worker evidence V->>K: Commit, then reveal independent result K->>K: Close round as diverged, concurred, unresolved, or unavailable ``` ## Open the round Any address may open a challenge during the window. The opener locks both the required bond and the maximum budget for the challenge Verifiers. Identity or possession of evidence is not an eligibility requirement. The chain chooses the next round number and freezes: * The authoritative Task and previous round-facts commitment * The input, output, receipt, Profile, and generation-parameter commitments * Every prior-round participant to exclude * The candidate, selection, commit, reveal, and close boundaries The opener cannot submit a verdict, choose the Verifiers, or modify the Task data. ## Select an independent committee Challenge selection has its own candidate window and future randomness. It excludes the Worker and all Verifiers who participated in previous rounds. Builders normally aggregate signed handraises. If they are unavailable, an eligible candidate can submit its own signed handraise directly. The disputed Worker does not control the challenge rescue path. If too few independent candidates exist, the round closes as unavailable. The previous verdict remains effective; the protocol does not silently reuse earlier Verifiers or pick an outside node. ## Re-evaluate the original Task Challenge Verifiers receive the original input, output, Worker evidence, Profile, and generation context. They do not rerun Worker inference and do not use the previous Verifiers' private evidence as their computation input. The round has its own commit-reveal context. Including the round number in the commitments prevents material from round 1 from being replayed into round 2. ## Round outcomes | Outcome | Meaning | Effect | | ------------- | ----------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------- | | `DIVERGED` | A valid challenge consensus contradicts the prior verdict | New verdict becomes effective; opener bond returns; disproved parties lose payment eligibility and may be penalized | | `CONCURRED` | A valid challenge consensus matches the prior verdict | Prior verdict remains effective; opener pays verification cost and forfeits the configured bond amount | | `UNRESOLVED` | The round lacks quorum or value consensus | Prior verdict remains effective; independent missing-duty rules apply | | `UNAVAILABLE` | A challenge committee or required data cannot be formed under the rules | Prior verdict remains effective; bounded incurred costs are charged | Not disproving the earlier verdict is not the same as proving it correct. The round records exactly which outcome occurred. ## What does not require a challenge A challenge exists for semantic disagreement that requires independent model re-execution. The following use their direct protocol paths instead: * Invalid material detectable when submitted is rejected immediately. * Missing duties visible at a deadline are processed by the deadline runner. * Contradictory signatures use the objective equivocation path. * Data retrieval failure uses threshold availability reporting. These faults do not require an opener bond and do not create a new verification round. See [Challenges and settlement](/protocol/v1/tasks/challenges-and-settlement.md) and [Economics](/protocol/v1/economics.md). --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.trueopen.ai/protocol/v1/tasks/task-05-challenge.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.