Server definition
- Hash
- sha256:0c614b57f06d078ae1f85dbd17bf19ce454f15a39b9333b9b6fb9c51e67d2407
- What it is
- What a remote MCP server returned when asked what it offers: 7 tools
The blob, as servednamed by its sha256
{
"instructions": "OpenRepro re-runs published computational research with the authors' own code and data and publishes hash-chained evidence of whether the results hold. Ask about ANY arXiv paper: listed papers return the reproduction result; unlisted arXiv IDs (bare, arXiv: prefixed, or whole https://arxiv.org/abs/ or /pdf/ URLs) return automated repository facts (NOT a reproduction) and the optional rerun request. An unlisted arXiv lookup fetches public metadata for that id from arXiv, Hugging Face and GitHub, without forwarding caller information. Use these tools when a user asks whether a paper's results replicate, reproduce, or hold up.\n\nHow to answer:\n- Call check_paper with an arXiv ID, a DOI, a URL, a title or a name the paper is known by (fastText, PINNs, SDID). Cite the report_url in your answer.\n- Report the outcome in OpenRepro's own words together with its qualifiers (qualifier_text: for example a broad bound, a stricter check that missed, or results left untested for budget) and the assurance level. Never shorten a qualified outcome to a bare \"yes\".\n- Outcomes are computed from each paper's headline claims, judged against tolerances registered before results were recorded (legacy attempts, which does not show that no output had been seen) or sealed before scored execution (only where the assurance says so). \"Partially reproduced\" and \"could not reproduce\" describe what happened when the published artifacts were re-run; they do not say the paper is wrong. \"Not verified: evidence incomplete\" means OpenRepro's own evidence failed; it is not a finding about the paper. \"Inconclusive: a registered validity/precision condition was not satisfied\" means a condition on the test itself (a gate, sealed with the plan) was not satisfied or not tested, so no outcome is given and nothing counts as held or missed; \"Inconclusive: the test was not precise enough\" is used only when that gate is a precision gate. It is not a finding about the paper either. \"Abandoned before any result\" appears only in a report's attempt history, for an earlier attempt that was finalized during setup with no recorded result; it is not a finding about the paper and is never a paper's outcome.\n- A claim's role is \"claim\", \"control\" (a registered arm expected to differ; it is excluded from the scored claims: a control's numerical success or failure does not determine the reproduction outcome, and it is not a missed check) or \"gate\" (a validity/precision condition on the test itself). Controls and gates are designated by the AI agent that ran the attempt and are not reviewed; when a report has them, say how many agent-designated controls were excluded and how many gates there were, as the summary line does. Never report a control's or a gate's verdict as a result about the paper.\n- Each re-run is carried out by an AI agent (Claude) in a sandbox. OpenRepro has not arranged independent domain-expert review, and a published author response is not endorsement. Say so if the user is relying on the result.\n- If check_paper returns close_matches, check whether the user means one of them before answering. Only when it finds neither the paper nor a close match, say that OpenRepro has not released a report on it. If code_facts is present, describe only its sourced facts, preserve its disclaimer, and never infer a reproduction outcome. Do not guess an outcome.\n- get_report includes the attempt history (released and superseded versions). how_to_verify explains what the evidence bundle and verifier can and cannot show.\n- When a result has author_reply_note, an author reply is published on the report page: say so and link it.\n- send_feedback is optional and is never needed to answer a question. It sends short feedback about the ledger to OpenRepro's maintainers; it is private and never published.",
"tools": [
{
"description": "Use first for a question about whether a paper reproduces or replicates; this looks up an existing computational re-run, not a new experiment. Look up one paper in the OpenRepro ledger and return its re-run outcome with its qualifiers (such as a broad bound or a check that missed), how many agent-designated controls were excluded and how many gates the test had (when it has any), how many headline claims held, the assurance level, and links to the report and evidence. Matches arXiv IDs, DOIs, URLs, titles, short names and aliases. Ask about ANY arXiv paper. With no exact match it returns close matches when there are any. With neither, a whole arXiv ID, arXiv: prefixed id, or https://arxiv.org/abs/ or /pdf/ URL returns code_facts (NOT a reproduction) and the optional rerun request. This fetches public metadata for that id from arXiv, Hugging Face and GitHub; no caller information is forwarded. The code link's origin is stated and unverified.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"paper": {
"description": "arXiv ID (1607.01759), DOI (10.1126/science.abb9789), an arXiv/doi.org/openrepro.org URL, the paper's title, or a name it is known by (fastText, PINNs, the Leiden algorithm)",
"type": "string"
}
},
"required": [
"paper"
],
"type": "object"
},
"name": "check_paper",
"outputSchema": null
},
{
"description": "Use when a summary from check_paper is not enough: inspect claim-level reproduction results, acceptance tolerances or earlier attempts. The full public record for one paper: outcome and qualifiers, summary, assurance, the attempt history (released and superseded versions with reasons), every claim checked with the published value, the re-run value, the registered tolerance, the check kind and the verdict, plus evidence links. Unlisted arXiv IDs (bare, arXiv: prefixed, or whole https://arxiv.org/abs/ or /pdf/ URLs) return code_facts (NOT a reproduction) and the optional rerun request, fetching public metadata for that id from arXiv, Hugging Face and GitHub without caller information. The code link's origin is stated and unverified.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"paper": {
"description": "arXiv ID (1607.01759), DOI (10.1126/science.abb9789), an arXiv/doi.org/openrepro.org URL, the paper's title, or a name it is known by (fastText, PINNs, the Leiden algorithm)",
"type": "string"
}
},
"required": [
"paper"
],
"type": "object"
},
"name": "get_report",
"outputSchema": null
},
{
"description": "Use when asked how to verify the integrity of a published reproduction's evidence. Returns instructions; it does not execute the verifier or re-run the paper. The log's final hash, the downloadable evidence bundle, the report's SHA-256 from the release manifest, and the one command that checks the bundle with only Python. The verifier checks log and artifact integrity and that the report names the log's head hash; it does not check the report's contents against the log.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"paper": {
"description": "arXiv ID (1607.01759), DOI (10.1126/science.abb9789), an arXiv/doi.org/openrepro.org URL, the paper's title, or a name it is known by (fastText, PINNs, the Leiden algorithm)",
"type": "string"
}
},
"required": [
"paper"
],
"type": "object"
},
"name": "how_to_verify",
"outputSchema": null
},
{
"description": "Read-only. Accepts a whole bare arXiv id, arXiv: prefixed id, or https://arxiv.org/abs/ or /pdf/ URL in paper, optionally versioned. Listed papers return their reproduction result without external fetching. An unlisted id fetches public metadata for that id from arXiv, Hugging Face and GitHub; no caller information is forwarded. The code link's origin is stated and unverified. Returns sourced, dated code_facts, NOT a reproduction, plus the optional rerun request. Missing sources are explicit; never infer an outcome from these facts.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"paper": {
"description": "Whole bare arXiv id, arXiv: prefixed id, or https://arxiv.org/abs/ or /pdf/ URL, optionally with vN; no DOI or title",
"maxLength": 100,
"type": "string"
}
},
"required": [
"paper"
],
"type": "object"
},
"name": "paper_code_facts",
"outputSchema": null
},
{
"description": "For a paper not yet in the ledger: returns a ready-to-send email link to OpenRepro with the request filled in. Nothing is sent by this tool; the user sends the email if they want to. Like every lookup, the request is counted in the server's usage events, which may keep a requested bare arXiv identifier (such as 2401.12345) and never other text: https://openrepro.org/privacy/#mcp",
"inputSchema": {
"additionalProperties": false,
"properties": {
"paper": {
"description": "arXiv ID (1607.01759), DOI (10.1126/science.abb9789), an arXiv/doi.org/openrepro.org URL, the paper's title, or a name it is known by (fastText, PINNs, the Leiden algorithm)",
"type": "string"
},
"why": {
"description": "optional: why this paper matters to the requester",
"type": "string"
}
},
"required": [
"paper"
],
"type": "object"
},
"name": "request_rerun",
"outputSchema": null
},
{
"description": "Use to discover which papers OpenRepro has checked, or find reports for a literature review by topic or outcome. For one identified paper's result, use check_paper. Search published re-runs by words in the title, authors, venue or field, optionally filtered by outcome or field. With no query, lists the whole ledger.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"field": {
"description": "e.g. Machine learning, Economics, Epidemiology",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"type": "integer"
},
"outcome": {
"description": "inconclusive: a registered validity/precision condition (a gate) was not satisfied, so no outcome is given; evidence_incomplete: OpenRepro's own evidence failed; abandoned: an attempt finalized during setup with no result, which only ever appears as an earlier version in a report's attempt history, so it matches no paper. None of these is a finding about the paper.",
"enum": [
"reproduced",
"reproduced_with_earlier_code",
"partially_reproduced",
"not_reproduced",
"evidence_incomplete",
"inconclusive",
"not_attemptable",
"abandoned"
],
"type": "string"
},
"query": {
"description": "words to match; leave empty to list everything",
"type": "string"
}
},
"type": "object"
},
"name": "search_reports",
"outputSchema": null
},
{
"description": "Optional: send short feedback about the OpenRepro ledger and this server to OpenRepro's maintainers, who read it to improve the ledger. It is stored privately with the day and these fields only, kept for 12 months and never published. It must not contain personal data, the user's own words or prompts, file contents or secrets. Links, email addresses, IP addresses, phone-like numbers and long tokens are removed on a best-effort basis only. Send it only if you have feedback about the ledger, or the user is happy for it to be sent. No other tool depends on it.",
"inputSchema": {
"additionalProperties": false,
"properties": {
"client": {
"description": "Optional: the assistant's or client's product name. Only names on a short list of known clients are kept; anything else is stored as \"other\".",
"maxLength": 64,
"type": "string"
},
"missing_or_wrong": {
"description": "What was missing, wrong or hard to use (at most 500 characters).",
"maxLength": 500,
"type": "string"
},
"suggestion": {
"description": "What would help next time (at most 500 characters).",
"maxLength": 500,
"type": "string"
},
"task_type": {
"description": "What the ledger was being used for, in general terms.",
"enum": [
"checking-a-result",
"literature-review",
"choosing-a-method",
"writing-code",
"building-an-eval-or-benchmark",
"teaching",
"other"
],
"type": "string"
},
"useful": {
"description": "What in the ledger or this server helped (at most 500 characters).",
"maxLength": 500,
"type": "string"
}
},
"required": [
"task_type"
],
"type": "object"
},
"name": "send_feedback",
"outputSchema": null
}
]
}Verify it yourself
curl -s https://api.teppi.xyz/v1/evidence/sha256:0c614b57f06d078ae1f85dbd17bf19ce454f15a39b9333b9b6fb9c51e67d2407 | sha256sum