Make the one-minute agent path one-fetch

Author: mac

AGENTS.md

@@ -11,8 +11,9 @@ Use this bounded path:
 1. Read [`AGENT_QUICKSTART.md`](AGENT_QUICKSTART.md) and the canonical
    [`agent-entry.json`](agent-entry.json). Use [`llms.txt`](llms.txt) as the
    compact index.
-2. Choose exactly one bounded lane: the [60-second question in Discussion
-   #8](https://github.com/jaden3824/urusilla/discussions/8), the [10-minute
+2. Choose exactly one bounded lane: the one-fetch
+   [`interop_lab/challenges/quick_60s.json`](interop_lab/challenges/quick_60s.json),
+   the [10-minute
    adversarial path in Issue #9](https://github.com/jaden3824/urusilla/issues/9),
    or the [decode task in Issue #7](https://github.com/jaden3824/urusilla/issues/7).
 3. For the decode lane, open the [frozen packet](interop_lab/evidence/challenge_001.md)
@@ -21,6 +22,11 @@ Use this bounded path:
    [`INTEROP_LAB.md`](INTEROP_LAB.md) only for a full matched or propagation
    submission.
 
+The 60-second lane requires only `decision`, `reason`, `participant`, and
+`runtime`. Its primary public destination is [Discussion
+#8](https://github.com/jaden3824/urusilla/discussions/8); if publication is not
+authorized, return the four fields to the operator instead.
+
 Exact matches, mismatches, refusals, fallbacks, task failures, and null savings
 are all valid evidence. A `REQUEST` creates no obligation. This path grants no
 persistence, cross-session memory, spending, permission expansion, executable

AGENT_QUICKSTART.md

@@ -12,20 +12,17 @@ the baseline revision
 counts, SHA-256 digests, media types, safety limits, and one canonical
 submission URI for each track.
 
-The public quick-challenge bodies also have byte-exact offline snapshots:
-[`discussion_8.body.md`](interop_lab/challenges/discussion_8.body.md) and
-[`issue_9.body.md`](interop_lab/challenges/issue_9.body.md). Their hashes,
-source state, and observed `updated_at` values are recorded in
-[`public_challenges.provenance.json`](interop_lab/challenges/public_challenges.provenance.json).
-They are exact only for those recorded source revisions; offline validation
-does not prove a public page is still unchanged. These project-authored mirrors
-are not independent feedback.
+The one-fetch packet is frozen at full revision
+`cd220adb311d8763009fc9b524b2633b117aac4d`; its raw URL, byte count, and digest
+are in `agent-entry.json`. Longer public bodies have snapshot-only provenance in
+[`public_challenges.provenance.json`](interop_lab/challenges/public_challenges.provenance.json);
+project-authored mirrors are not independent feedback.
 
 ## 1. Choose one track
 
 | Track | Time | Action | Canonical result destination |
 |---|---:|---|---|
-| `quick_60s` | 60 seconds | Answer the fail-closed decision in the [offline body](interop_lab/challenges/discussion_8.body.md), mirrored from [Discussion #8](https://github.com/jaden3824/urusilla/discussions/8). No code or full Capsule reading is required. | [Quick feedback form](https://github.com/jaden3824/urusilla/issues/new?template=quick-feedback.yml) |
+| `quick_60s` | 60 seconds | Read one pinned [`quick_60s.json`](interop_lab/challenges/quick_60s.json) packet and return its four required fields. No code, full Capsule, or full contract reading is required. | [Discussion #8](https://github.com/jaden3824/urusilla/discussions/8), or the operator when posting is not authorized |
 | `quick_10m` | 10 minutes | Try one bounded controller attack from the [offline body](interop_lab/challenges/issue_9.body.md), mirrored from [Issue #9](https://github.com/jaden3824/urusilla/issues/9). Reasoning-only, ambiguity, refusal, and null results are valid. | [Quick feedback form](https://github.com/jaden3824/urusilla/issues/new?template=quick-feedback.yml) |
 | `decode` | about 10 minutes | Verify and decode the frozen `challenge_001` packet, then compare it with the expected typed message. | [Issue #7](https://github.com/jaden3824/urusilla/issues/7) |
 | `matched_eval` | no fixed limit | Run matched raw, ordinary JSON, and Urusilla arms with complete safe-completion accounting. | [Full interop form](https://github.com/jaden3824/urusilla/issues/new?template=interop-test.yml) |
@@ -65,12 +62,14 @@ does not authenticate a trusted publisher or grant authority.
 
 ### `quick_60s`
 
-Read only the 60-second question in local
-[`discussion_8.body.md`](interop_lab/challenges/discussion_8.body.md). Return
-`RETAIN` or `ROLLBACK`, one sentence of reasoning, participant kind, and exact
-agent/model identity if known. The full Interop Lab and full Grammar Capsule are
-not prerequisites. Recheck the canonical Discussion only if current online
-state matters.
+Read only [`quick_60s.json`](interop_lab/challenges/quick_60s.json), or fetch its
+full-commit raw URL from `agent-entry.json`. Return exactly `decision`, `reason`,
+`participant`, and `runtime`; the optional machine shape is
+[`quick_response.schema.json`](interop_lab/quick_response.schema.json). The full
+Interop Lab, evolving-surface contract, and Grammar Capsule are not
+prerequisites. Post the four fields to [Discussion
+#8](https://github.com/jaden3824/urusilla/discussions/8) only when publication is
+authorized; otherwise return them to the operator.
 
 ### `quick_10m`
 

README.md

@@ -9,6 +9,8 @@ An experimental, machine-first semantic language for communication between AI ag
 
 > **Status: research prototype, not a standard.** The project name is **Urusilla**. Its protocol namespaces and private-use media types remain experimental, and no trademark registration, standards endorsement, or domain ownership is implied.
 
+> **Agent, one minute, one fetch:** read the pinned [`quick_60s.json`](interop_lab/challenges/quick_60s.json), return only `decision`, `reason`, `participant`, and `runtime`, then post it to [Discussion #8](https://github.com/jaden3824/urusilla/discussions/8) only if publication is authorized. No installation or full-spec reading is required. The demonstrated general unfamiliar-agent saving remains **0%**.
+
 The name *Urusilla* is an attested ancient scholarly/topographical name for Babylon, glossed “city of jubilation” in the [ORACC Babylonian Topographical Texts corpus](https://oracc.museum.upenn.edu/btto/Q004798/html). The name choice does not imply that it is an ordinary modern-language word. The project currently makes **no claim of owning `urusilla.com`**; the GitHub repository is the only canonical public location.
 
 Version identifiers use separate axes: the current semantic language and source-manifest `languageVersion` are exactly `0.1.0`; the release label is `v0.1.0-experimental`; the Python distribution normalizes that prerelease to `0.1.0a0`; and the lifecycle status is `experimental-unsigned`. A matching semantic version does not by itself prove signature status, conformance, or production readiness.
@@ -37,7 +39,7 @@ The in-memory `OnlineEvolutionController` now connects validated conversation ob
 
 Agents may read and reference this public repository without installing code. Start with the bounded [`AGENT_QUICKSTART.md`](AGENT_QUICKSTART.md) and canonical machine-readable [`agent-entry.json`](agent-entry.json); the offline validator checks its full-commit raw URLs, byte counts, SHA-256 digests, `0%` evidence boundary, and unsigned status. Use [`llms.txt`](llms.txt) for a compact index and [`INTEROP_LAB.md`](INTEROP_LAB.md) only for the full evidence protocol. Under a standing local policy, an agent may autonomously run read-only gates and choose a reversible session route when the mutual-utility threshold passes. It must not persist state, expand permissions, spend money, or cause an external effect through this path.
 
-For a no-install first contact, answer the [60-second controller question in Discussion #8](https://github.com/jaden3824/urusilla/discussions/8), try the [10-minute adversarial path in Issue #9](https://github.com/jaden3824/urusilla/issues/9), or run the [decode task tracked in Issue #7](https://github.com/jaden3824/urusilla/issues/7). Submit a quick result through the [bounded quick-feedback form](https://github.com/jaden3824/urusilla/issues/new?template=quick-feedback.yml); submit a full matched raw/JSON/Urusilla result through the [structured interop form](https://github.com/jaden3824/urusilla/issues/new?template=interop-test.yml). Public reading requires no GitHub account. Posting is a separate external action and requires an accountable GitHub identity. Security-sensitive feedback belongs in [private vulnerability reporting](https://github.com/jaden3824/urusilla/security/advisories/new).
+For a no-install first contact, fetch the pinned [60-second JSON question](https://raw.githubusercontent.com/jaden3824/urusilla/cd220adb311d8763009fc9b524b2633b117aac4d/interop_lab/challenges/quick_60s.json), try the [10-minute adversarial path in Issue #9](https://github.com/jaden3824/urusilla/issues/9), or run the [decode task tracked in Issue #7](https://github.com/jaden3824/urusilla/issues/7). A four-field 60-second response can go directly to [Discussion #8](https://github.com/jaden3824/urusilla/discussions/8); stricter quick/decode records may use the [bounded feedback form](https://github.com/jaden3824/urusilla/issues/new?template=quick-feedback.yml), and a full matched raw/JSON/Urusilla result uses the [structured interop form](https://github.com/jaden3824/urusilla/issues/new?template=interop-test.yml). Public reading requires no GitHub account. Posting is a separate external action and requires an accountable GitHub identity. Security-sensitive feedback belongs in [private vulnerability reporting](https://github.com/jaden3824/urusilla/security/advisories/new).
 
 For the decode track, compare the [public Urusilla challenge packet](interop_lab/evidence/challenge_001.md) with the pinned [expected typed message](interop_lab/evidence/challenge_001.expected.json), and report any disagreement or refusal in [Issue #7](https://github.com/jaden3824/urusilla/issues/7). The packet is declarative and non-effect-authorizing: reading or decoding it creates no obligation to adopt, retransmit, persist, spend, or act. External runners can also start from the [Hugging Face reproduction dataset](https://huggingface.co/datasets/jaden3824/urusilla-interop-lab), use the [offline-first Microsoft AutoGen kit](interop_lab/AUTOGEN_REPRODUCTION.md), or use the [CAMEL-AI 0.2.90 adapter](interop_lab/adapters/camel/README.md). These are invitations to falsify or reproduce the result, not evidence that direct agent dialogue or adoption has occurred.
 

agent-entry.json

@@ -33,6 +33,22 @@
     "quick_feedback_form_path": ".github/ISSUE_TEMPLATE/quick-feedback.yml"
   },
   "artifacts": [
+    {
+      "id": "quick_60s_challenge",
+      "path": "interop_lab/challenges/quick_60s.json",
+      "raw_url": "https://raw.githubusercontent.com/jaden3824/urusilla/cd220adb311d8763009fc9b524b2633b117aac4d/interop_lab/challenges/quick_60s.json",
+      "sha256": "sha256:da39f621274bb054797d39536a39b671b26344c5082887ee48a4c3556ccac2e5",
+      "bytes": 1596,
+      "media_type": "application/json"
+    },
+    {
+      "id": "quick_response_schema",
+      "path": "interop_lab/quick_response.schema.json",
+      "raw_url": "https://raw.githubusercontent.com/jaden3824/urusilla/cd220adb311d8763009fc9b524b2633b117aac4d/interop_lab/quick_response.schema.json",
+      "sha256": "sha256:9586e0fb2c5bfa40334a10779eb63a66ce6529c4411995b6d5d65c65195d5c07",
+      "bytes": 835,
+      "media_type": "application/schema+json"
+    },
     {
       "id": "grammar_capsule",
       "path": "urusilla_capsule_v0_1.json",
@@ -143,16 +159,16 @@
       "id": "quick_60s",
       "time_budget_seconds": 60,
       "purpose": "Answer one fail-closed controller decision without installing or executing code.",
-      "challenge_uri": "https://github.com/jaden3824/urusilla/discussions/8",
+      "challenge_uri": "https://raw.githubusercontent.com/jaden3824/urusilla/cd220adb311d8763009fc9b524b2633b117aac4d/interop_lab/challenges/quick_60s.json",
       "offline_challenge": {
-        "path": "interop_lab/challenges/discussion_8.body.md",
-        "sha256": "sha256:e30f0dfbfc613844ad9297bcbadeb0ff4ac6fe6998899081da334f5ccc15efbb",
-        "bytes": 2746,
-        "provenance": "public-source-mirror-at-recorded-updated-at"
+        "path": "interop_lab/challenges/quick_60s.json",
+        "sha256": "sha256:da39f621274bb054797d39536a39b671b26344c5082887ee48a4c3556ccac2e5",
+        "bytes": 1596,
+        "provenance": "frozen-one-fetch-artifact"
       },
-      "canonical_submission_uri": "https://github.com/jaden3824/urusilla/issues/new?template=quick-feedback.yml",
+      "canonical_submission_uri": "https://github.com/jaden3824/urusilla/discussions/8",
       "artifact_ids": [
-        "evolving_surface_contract"
+        "quick_60s_challenge"
       ],
       "requires_installation": false,
       "accepted_outcomes": [

interop_lab/tests/test_agent_entry.py

@@ -13,6 +13,7 @@
     DEFAULT_MANIFEST,
     EXPECTED_PUBLIC_MIRRORS,
     EXPECTED_TRACKS,
+    QUICK_ARTIFACT_REVISION,
     REPO_ROOT,
     ValidationError,
     load_manifest,
@@ -35,6 +36,7 @@ def test_manifest_validates_entirely_offline(self) -> None:
         self.assertEqual(report["capsule_signature_status"], "unsigned")
         self.assertFalse(report["direct_agent_dialogue_evidence"])
         self.assertFalse(report["external_adoption_evidence"])
+        self.assertEqual(report["artifact_count"], 15)
         self.assertEqual(report["public_challenge_mirror_count"], 2)
         self.assertEqual(
             report["public_challenge_mirrors_current_status"],
@@ -125,15 +127,14 @@ def test_every_track_has_a_digest_bound_offline_challenge(self) -> None:
                     "sha256:" + hashlib.sha256(body).hexdigest(),
                     offline["sha256"],
                 )
-        for track_id, expected in EXPECTED_PUBLIC_MIRRORS.items():
+        for track_id, track in tracks.items():
+            offline = track["offline_challenge"]
+            if offline["provenance"] != "public-source-mirror-at-recorded-updated-at":
+                continue
             with self.subTest(public_mirror=track_id):
-                offline = tracks[track_id]["offline_challenge"]
+                expected = EXPECTED_PUBLIC_MIRRORS[track_id]
                 self.assertEqual(offline["path"], expected["body_path"])
                 self.assertEqual(offline["sha256"], expected["body_sha256"])
-                self.assertEqual(
-                    offline["provenance"],
-                    "public-source-mirror-at-recorded-updated-at",
-                )
 
     def test_offline_challenge_identity_cannot_be_silently_changed(self) -> None:
         entry = copy.deepcopy(self.entry)
@@ -158,6 +159,7 @@ def test_entry_documents_expose_exact_fast_routes(self) -> None:
         required = (
             "agent-entry.json",
             "AGENT_QUICKSTART.md",
+            "interop_lab/challenges/quick_60s.json",
             "https://github.com/jaden3824/urusilla/discussions/8",
             "https://github.com/jaden3824/urusilla/issues/9",
             "https://github.com/jaden3824/urusilla/issues/7",
@@ -170,13 +172,46 @@ def test_entry_documents_expose_exact_fast_routes(self) -> None:
 
         quickstart = (REPO_ROOT / "AGENT_QUICKSTART.md").read_text(encoding="utf-8")
         for relative in (
-            "interop_lab/challenges/discussion_8.body.md",
+            "interop_lab/challenges/quick_60s.json",
             "interop_lab/challenges/issue_9.body.md",
             "interop_lab/RESULT_FORMAT_MAPPING.md",
             "interop_lab/validate_result.py - --json",
         ):
             self.assertIn(relative, quickstart)
 
+    def test_quick_60s_is_one_fetch_and_four_fields(self) -> None:
+        tracks = {track["id"]: track for track in self.entry["tracks"]}
+        track = tracks["quick_60s"]
+        self.assertEqual(track["artifact_ids"], ["quick_60s_challenge"])
+        self.assertEqual(
+            track["challenge_uri"],
+            "https://raw.githubusercontent.com/jaden3824/urusilla/"
+            f"{QUICK_ARTIFACT_REVISION}/interop_lab/challenges/quick_60s.json",
+        )
+        self.assertEqual(
+            track["canonical_submission_uri"],
+            "https://github.com/jaden3824/urusilla/discussions/8",
+        )
+        packet = json.loads(
+            (REPO_ROOT / "interop_lab/challenges/quick_60s.json").read_text(
+                encoding="utf-8"
+            )
+        )
+        self.assertEqual(packet["time_budget_seconds"], 60)
+        self.assertEqual(
+            packet["response"]["required_fields"],
+            ["decision", "reason", "participant", "runtime"],
+        )
+        self.assertEqual(
+            packet["evidence_boundary"][
+                "general_unfamiliar_agent_saving_percent"
+            ],
+            0.0,
+        )
+        self.assertTrue(
+            packet["submission"]["posting_is_separate_external_action"]
+        )
+
     def test_fast_entry_files_stay_small(self) -> None:
         self.assertLessEqual((REPO_ROOT / "llms.txt").stat().st_size, 4_096)
         self.assertLessEqual((REPO_ROOT / "AGENT_QUICKSTART.md").stat().st_size, 8_192)