skilldrop-cli 0.16.28 → 0.16.30

package/packs/research/loops/research/evals/evals.json

@@ -0,0 +1,39 @@
+{
+  "loop_name": "research",
+  "evals": [
+    {
+      "id": "plans-before-reading",
+      "prompt": "Run the research loop: why did our trial-to-paid conversion drop from 14% to 9% in September? I've put the sources in sources/ (a product analytics export, the September release notes, 12 sales call notes, and a competitor's pricing page). The growth lead decides by 1 November whether to roll back the new onboarding flow.",
+      "assertions": [
+        "Runs research-plan first and writes, before synthesising, what finding would change the roll-back decision",
+        "States the stopping rule before reading the sources",
+        "The synthesis cites every claim to a listed source and location",
+        "The comparison includes a measurement hypothesis (tracking or definition change) alongside the onboarding flow and the competitor's pricing",
+        "Every evidence row in the matrix cites a listed source or is marked [assumption]",
+        "Ends with a G7 verdict from PROCEED, PROCEED WITH CONDITIONS, REVISE, RECONSIDER or BLOCKED, and the reasons for it",
+        "The reply leads with the explanation the evidence favours and what it means for the roll-back, not with the matrix"
+      ]
+    },
+    {
+      "id": "revise-brings-the-discriminating-observation",
+      "prompt": "Continuing the research loop. The comparison left H1 (the new onboarding flow) and H3 (the competitor's price cut on 4 September) tied, and named as discriminating: conversion for trials that started before the onboarding release on 11 September. I've added that cohort export as sources/cohorts.csv.",
+      "assertions": [
+        "Treats the tie as G7 REVISE and returns to synthesise, not to plan",
+        "Adds the cohort export as a new numbered source and re-synthesises with it cited",
+        "Re-runs the comparison with the new evidence row rated across every hypothesis",
+        "Counts this as round 2 under the cap of 3",
+        "Does not add further sources that both H1 and H3 predict equally"
+      ]
+    },
+    {
+      "id": "blocked-without-a-decision",
+      "prompt": "Run the research loop on why remote work is good. Non-interactive, no follow-up questions.",
+      "assertions": [
+        "Emits BLOCKED: need the decision this research informs, who makes it, and by when",
+        "Does not run source-synthesis or hypothesis-comparison",
+        "Does not invent a decision or gather sources",
+        "Says in plain words what it will produce once it has the decision"
+      ]
+    }
+  ]
+}

package/packs/research/loops/research/loop.json

@@ -0,0 +1,35 @@
+{
+  "$comment": "RFC-0052. A question to an answer the evidence can carry. The gate sits after the comparison because a synthesis can't tell you whether its own evidence separates the explanations; the matrix can.",
+  "name": "research",
+  "version": "0.1.0",
+  "description": "Take a question that turns on why something happened to an answer the evidence can carry: a plan tied to the decision it serves, a synthesis of the sources with every claim cited, and rival explanations ranked by how little evidence contradicts them, with a gate that sends the work back for the observation that would tell the leaders apart. Use when the user says \"research why…\", \"find out what's causing…\", \"run the research loop\", or needs a decision-ready answer on which explanation the evidence supports.",
+  "entrypoint": "LOOP.md",
+  "kind": "loop",
+  "cap": 3,
+  "stages": [
+    {
+      "id": "plan",
+      "type": "generate",
+      "intent": "Pin the decision the research serves, the precise question, what finding would change the decision, and when to stop, before any source is read.",
+      "skills": ["research-plan"]
+    },
+    {
+      "id": "synthesise",
+      "type": "generate",
+      "intent": "Say what the sources gathered under the plan support, every claim cited to a location, a repeated source counted once.",
+      "skills": ["source-synthesis"]
+    },
+    {
+      "id": "compare",
+      "type": "verify",
+      "intent": "Rank the rival explanations against the cited evidence, then judge whether that evidence is strong enough to carry the decision.",
+      "skills": ["hypothesis-comparison"],
+      "gate": {
+        "id": "G7",
+        "kind": "review",
+        "verdicts": ["PROCEED", "PROCEED WITH CONDITIONS", "REVISE", "RECONSIDER", "BLOCKED"],
+        "revise_to": "synthesise"
+      }
+    }
+  ]
+}