{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_t42t43jdsbqj","handle":"touchstone-mcp-2","url":"https://wellknown.network/agents/touchstone-mcp-2","links":{"self":"https://wellknown.network/agents/touchstone-mcp-2/record.json","html":"https://wellknown.network/agents/touchstone-mcp-2","markdown":"https://wellknown.network/agents/touchstone-mcp-2/record.md","api":"https://wellknown.network/api/v1/agents/touchstone-mcp-2","status":"https://wellknown.network/api/v1/agents/touchstone-mcp-2/status","claim":"https://wellknown.network/agents/touchstone-mcp-2/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/touchstone-mcp-2/claim.json","badge":"https://wellknown.network/agents/touchstone-mcp-2/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/touchstone-mcp-2/history","tools":"https://wellknown.network/api/v1/agents/touchstone-mcp-2/tools"},"ard":{"identifier":"urn:air::server:touchstone-mcp-2","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"touchstone-mcp","summary":"A cheap, deterministic first-pass filter for unsupported claims in LLM output, computed without a second model call. Not a standalone hallucination detector. MCP server plus the reference implementation of Touchstone Standard 1.0.","description":"# Touchstone\n\nA cheap, deterministic first-pass filter for unsupported claims in LLM\noutput. Not a standalone hallucination detector.\n\nTouchstone scores whether LLM-generated text is supported by its source\nusing regex, structural analysis, and arithmetic, with no second model\ncall. You get a probability, a flag for when the input is too thin to\nscore reliably, and highlights of the sentences that look unsupported.\n\nRead this first: Touchstone is structurally blind to hallucinations that\nkeep the source vocabulary and only change the meaning. On a 16-case\nsubtle-hallucination stress test it separates at 50%, no better than\nrandom. Where it earns its keep is triage (around 4x lift over random\nreview), a cheap first stage ahead of an LLM judge, and drift detection.\nSee `docs/production_readiness.md` before relying on it.\n\n## Quickstart (60 seconds)\n\n```bash\npip install touchstone-mcp\n```\n\n```python\nfrom clarethium_touchstone import Verifier\n\nv = Verifier()\nresult = v.score(\n    text=(\n        \"Apple reported Q1 fiscal 2026 revenue of $185 billion, the company's \"\n        \"highest ever. McKinsey forecasts industry-wide growth of 47% next \"\n        \"quarter. The Federal Reserve will raise rates 75 basis points in \"\n        \"response. Tesla announced a competing AR product for late 2027.\"\n    ),\n    source=(\n        \"Apple reported Q1 fiscal 2026 revenue of $143 billion. The company's \"\n        \"iPhone segment grew 8% year-over-year. Tim Cook commented on AI \"\n        \"investments during the earnings call. Operating margins reached 32%.\"\n    ),\n)\nprint(f\"prob_hallucinated = {result.prob_hallucinated:.3f}\")  # → 0.796\nprint(f\"scope             = {result.scope}\")                  # → validated\nprint(f\"should_flag()     = {result.should_flag()}\")          # → True\nfor span in result.top_unsupported:\n    print(f\"  [{span.layer11_primary}] {span.sentence!r} {span.p_markers}\")\n```\n\nWhat just happened: Touchstone flagged three sentences as projected\n(unsupported), the unsource…","publisher":null,"homepage":"https://github.com/lluvr/touchstone/tree/main/docs","repository":"https://github.com/lluvr/touchstone/blob/main/CHANGELOG.md","version":"0.1.11","license":null,"protocols":["mcp"],"tags":["ai","ai-evaluation","ai-output","fabrication-detection","grounding","hallucination-detection","mcp","model-context-protocol","structural-verification","touchstone","clarethium","model-independent"],"pricing":null,"endpoints":[{"url":"pypi:touchstone-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","repoUrl":"pypi","summary":"pypi","version":"pypi","description":"pypi","homepageUrl":"pypi"}},"derived":{"capabilities":[{"slug":"ai.evaluation","name":"Evaluation & Benchmarks","confidence":1,"provenance":"derived"},{"slug":"dev.docs-lookup","name":"Documentation Lookup","confidence":0.791,"provenance":"derived"}],"categories":["ai","dev"],"language":"en"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"touchstone-mcp","registry":"pypi","observedAt":"2026-10-29T00:27:13.909Z","publishedAt":"2026-07-03T09:06:53.821582Z","latestVersion":"0.1.11"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"touchstone-mcp","url":"https://pypi.org/project/touchstone-mcp/","firstSeenAt":"2026-09-10T14:22:59.461Z","fetchedAt":"2026-10-29T00:25:59.021Z","normalizedAt":"2026-10-29T00:25:59.021Z"}]},"firstSeenAt":"2026-09-10T14:22:59.461Z","updatedAt":"2026-10-29T00:27:13.909Z"}