Merge pull request #17: Extract automatic Planner research methodology

README.md

@@ -99,3 +99,5 @@ communication treatment is off. Unrelated automatic packages remain available.
 The owner retains publication identity/order, mandatory status, complete durable
 artifacts, verification and Operator acceptance. These packages customize
 answer/detail/example/acknowledgement presentation only.
+
+The automatic `research-auto` Planner package supplies primary-source selection, preferred-domain exploration/refinement and evidence-reading methodology, publication/version checking, corroboration and already-admitted empirical evidence methods. The manual `research` package remains available on request. Tool arguments, budgets, continuation, recovery and untrusted research provenance belong to the owner.

packages/research-auto/SKILL.md

@@ -0,0 +1,23 @@
+---
+name: research-auto
+description: Source selection and reading methods for assigned Planner research.
+metadata:
+  relayview-display-name: Research methodology
+  relayview-version: 0.12.2
+---
+
+# Research methodology
+
+Prefer primary sources: applicable official documentation, specifications, source code, release notes, maintainer records, and reproducible first-party artifacts that own a claim; use secondary sources for orientation and corroboration. When an applicable authoritative domain is already known, preferredDomains may rank it higher; omit the option during broad exploration and refine a later search after discovering the right domain. Prefer narrow publisher hosts over broad multi-tenant hosts, and inspect/fetch returned sources before relying on them.
+
+When external research is material, use web_search to discover candidate public sources before web_fetching selected applicable evidence instead of guessing a destination URL.
+
+Check publication date, applicable product and version, and whether a source supports the precise claim. State uncertainty, conflicts, missing access, and inferences explicitly.
+
+Corroborate consequential or surprising claims with independent evidence when practical; mirrors, syndication, repetition, and pages derived from one upstream source are not independent.
+
+For an empirically testable technical claim, prefer a small reproducible local experiment only when this role already has an approved execution capability. Record the question, procedure, relevant versions and resolved paths, observation, and limitations; do not generalize beyond the tested environment.
+
+Use confidence language proportional to the evidence and identify claims that still depend on provider ranking, inaccessible sources, or unverified assumptions.
+
+Repeat an exact query only when the cached result is intentionally sufficient; use a materially narrower query when new terms, version, claim, or source scope justify a distinct auditable search.

packages/research-auto/relayview-skill.json

@@ -0,0 +1,15 @@
+{
+  "schemaVersion": 1,
+  "source": {
+    "kind": "first-party",
+    "reference": "v0.12.2"
+  },
+  "role": "planner",
+  "step": "planning",
+  "files": [
+    {
+      "path": "SKILL.md",
+      "sha256": "112036daeb1f760093345805b8c17b1f385723f3597608cf8ad19d72023ccd74"
+    }
+  ]
+}

tests/test_packages.py

@@ -12,6 +12,17 @@ ROOT = Path(__file__).resolve().parents[1]
 
 
 class PackageTests(unittest.TestCase):
+    def test_research_auto_contains_only_extracted_method(self):
+        package = ROOT / "packages/research-auto"
+        check_package(package)
+        text = (package / "SKILL.md").read_text()
+        self.assertIn("omit the option during broad exploration", text)
+        self.assertIn("inspect/fetch returned sources", text)
+        self.assertIn("Check publication date", text)
+        self.assertIn("Corroborate consequential", text)
+        self.assertNotIn("spawn", text)
+        self.assertNotIn("save a research file", text)
+
     def test_communication_packages_declare_their_treatment(self):
         for role in ("planner", "implementer", "reviewer"):
             package = ROOT / f"packages/{role}-communication"