{"id":15264,"plugin_id":"plugin_asdk_app_6aa9463431d88191a4fbb11f033b2d43","kind":"skill","collection_source":null,"comparison_source":null,"observed_at":"2026-09-30T23:11:05.896Z","digest":"3060ba4414af122f2600188590cdb384793426a3bf191147c0cf8fbacff85221","against":null,"payload":{"name":"falcon-verify","description":"Check whether an AI-written answer is supported by the evidence it was written from, using FALCON Verify's falcon_verify tool. Use when a user asks \"is this supported\", wants a summary checked against a document or data, needs every unsupported claim listed, or wants an auditable record of the check. Not for checking whether something is true in the world; FALCON verifies against the evidence supplied, nothing else.","included_files":[],"skill_md_contents":"---\nname: falcon-verify\ndescription: Check whether an AI-written answer is supported by the evidence it was written from, using FALCON Verify's falcon_verify tool. Use when a user asks \"is this supported\", wants a summary checked against a document or data, needs every unsupported claim listed, or wants an auditable record of the check. Not for checking whether something is true in the world; FALCON verifies against the evidence supplied, nothing else.\n---\n\n# FALCON Verify\n\nFALCON Verify answers one question: is this output supported by this evidence?\nIt does not know the world. A claim the evidence cannot speak to comes back\n`unverifiable`, never `supported`. Every call leaves a hash-chained audit record.\n\n## When to call\n\n- The user has an output (a summary, an answer, a report, a chatbot reply) AND\n  the material it was written from (a document, a record, a set of facts, a\n  retrieved passage). Call `falcon_verify`.\n- The user asks whether a run has settled, or for its journal: call\n  `falcon_verify_status` or `falcon_get_journal` with the run id.\n- The user asks how many verifications are left this month: call `falcon_account`.\n- The user only has a statement and no evidence: do not call. Say that FALCON\n  verifies against evidence the user supplies, and ask for it.\n\n## How to call `falcon_verify`\n\n- `output`: the prose to verify, verbatim.\n- `evidence`: an object, one key per source, each value the source text or data\n  VERBATIM AND IN FULL. Never summarise or paraphrase the evidence: the verdicts\n  stand against exactly what is passed, and condensed evidence makes true claims\n  read unsupported. A pasted document is fine as `{\"document\": \"<text>\"}`.\n- `claims` (optional): when the user has already listed the claims, pass each\n  as an object `{\"text\": \"<the output's own words>\", \"type\": \"<numeric |\n  comparative | temporal | categorical | causal | existential>\"}`. Never pass\n  bare strings. With claims supplied, no model call is needed to find them.\n- `depth`: leave at the default `auto` unless the user asks for rules only\n  (`extract`) or insists on a judge for every claim (`judge`, about three\n  verifications' cost for one).\n\n## Reading the result\n\n- `status` is the epistemic status of the check, not the verdict:\n  `supported` (every decided claim held), `qualified` (some did not),\n  `insufficient_support` (nothing could be decided), `execution_failure`.\n- `claims[]` carries one entry per claim with its `verdict` and the evidence\n  fields it was bound to. Report each claim by its own words and verdict.\n- `unsupported_rate` is over DECIDED claims and is `null` when none were\n  decided. A null rate is not a clean result; say what could not be checked.\n- `insufficiency`, when present, says what was established, what was not, why,\n  and what evidence or depth would resolve each item. Relay it.\n- `attribution` and `report_as` give the line to close with (FALCON Verify, the\n  run id, the counts). Use it as written.\n- `audit.run_id` is what the user needs to fetch the journal or status later.\n\n## What not to do\n\n- Do not present a FALCON verdict as a statement about the world. \"Supported\"\n  means supported by the evidence given.\n- Do not invent or complete evidence. If the user's evidence is partial, verify\n  against it and say what was missing.\n- Do not offer prices, plans or purchase links. If the allowance is exhausted,\n  the tool says so; tell the user their Account page on falcon.1-i.ai shows\n  their plan, and stop there.\n\n## Example\n\nUser: \"Verify: 'Q2 revenue was $4.2M, up 12% year over year, and gross margin\nreached 31%.' Facts: 'Second-quarter revenue came in at $4.2 million, a 12%\nincrease over the same quarter last year. Gross margin was 30.6%.'\"\n\nCall `falcon_verify` with `output` set to the quoted sentence and `evidence`\nset to `{\"facts\": \"<the facts text verbatim>\"}`. Report: revenue $4.2M\nsupported; up 12% supported; margin 31% unsupported (the evidence says 30.6%).\nClose with the attribution line from the result.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}