diff --git a/context7.json b/context7.json
index 3d0558d..b8d4252 100644
--- a/context7.json
+++ b/context7.json
@@ -1,18 +1,30 @@
 {
   "$schema": "https://context7.com/schema/context7.json",
+  "url": "https://context7.com/sonilo-ai/sonilo-python",
+  "public_key": "pk_h4BOrwVwv5CF9NJ5BIwvC",
   "projectTitle": "Sonilo",
   "description": "Official Python client and CLI for the Sonilo API — generate music, sound effects, and combined soundtracks from text or video.",
-  "excludeFolders": ["**/dist/**", "**/.venv/**", "**/__pycache__/**", ".superpowers", ".pytest_cache"],
+  "excludeFolders": [
+    "**/dist/**",
+    "**/.venv/**",
+    "**/__pycache__/**",
+    ".superpowers",
+    ".pytest_cache"
+  ],
   "rules": [
-    "Read the API key from the SONILO_API_KEY environment variable by default; only pass api_key= explicitly when the caller has a reason to override it.",
-    "For text-to-music and video-to-music: use client.text_to_music.generate() / client.video_to_music.generate() (streaming) for short synchronous tracks. Switch to generate_async() (or submit() + client.tasks.wait(parser=parse_music_result)) when the caller needs output_format=\"wav\" — and, for video-to-music only, isolate_vocals, preserve_speech, or ducking. The streaming path supports none of these.",
-    "For text-to-sfx and video-to-sfx, generation is always async: submit() returns a task, and generate() (or tasks.wait()) polls it to completion. There is no streaming variant for these two endpoints.",
-    "client.video_to_sound and client.video_to_video_sound score one clip with a music bed AND sound effects in a single call — prefer them over chaining video-to-music with video-to-sfx, which costs two charges. Both are async-only and take identical options: music_prompt and sfx_prompt (NOT a single prompt), preserve_speech, and ducking. video_to_sound returns mixed audio (output_type \"audio\"); video_to_video_sound returns the source video with that audio muxed in (output_type \"video\").",
-    "ducking is default-ON server-side and the request builder only sends a boolean when it is not None. Leave ducking unset (None) to keep the default; pass ducking=False ONLY to explicitly opt out. Never forward a default of False — that silently disables the server default. The same tri-state applies to preserve_speech.",
-    "A SoundResult exposes the combined render as the bare presigned output_url — save it with result.save(path). The separate layers come back as the music, music_processed, and sfx stems, saved with result.save_stem(path, which=\"music\"). music_processed is present only when preserve_speech or ducking altered the music bed, and save_stem raises SoniloError for an absent stem.",
+    "Read the API key from the SONILO_API_KEY environment variable by default; pass api_key= explicitly only when the caller has a reason to override it.",
+    "text-to-music and video-to-music: use generate() (streaming) for short synchronous tracks. It does NOT support output_format=\"wav\", isolate_vocals, preserve_speech or ducking.",
+    "For those options use generate_async(), or submit() + client.tasks.wait(parser=parse_music_result). isolate_vocals/preserve_speech/ducking are video-to-music only; text-to-music takes output_format alone.",
+    "text-to-sfx and video-to-sfx are always async: submit() returns a task and generate() (or tasks.wait()) polls it to completion. Neither has a streaming variant.",
+    "client.video_to_sound and client.video_to_video_sound score one clip with a music bed AND sound effects in a single call — prefer them over chaining video-to-music with video-to-sfx, which is charged twice.",
+    "Both sound endpoints are async-only and take identical options: music_prompt and sfx_prompt (NOT a single prompt), preserve_speech, ducking. video_to_sound returns mixed audio; video_to_video_sound returns the video with that audio muxed in.",
+    "ducking is default-ON server-side and the request builder only sends a boolean when it is not None. Leave it unset to keep the default; pass ducking=False only to opt out. Never forward a default of False — that silently disables it.",
+    "The same tri-state applies to preserve_speech: None means unset, not False.",
+    "SoundResult exposes the combined render as the presigned output_url — save it with result.save(path). Individual layers come from result.save_stem(path, which=\"music\"), with stems music, music_processed and sfx.",
+    "music_processed exists only when preserve_speech or ducking altered the music bed; save_stem raises SoniloError for an absent stem.",
     "Result media (.url) is a short-lived presigned URL, not the API's own domain — download it with the result's .save() helper; do not send the Authorization header to it.",
-    "Wrap calls in try/except for AuthenticationError (401), PaymentRequiredError (402), RateLimitError (429), and TaskFailedError (task reached status \"failed\") rather than a single generic except — callers usually want to handle these differently.",
-    "video / video_url parameters accept exactly one of the two, never both and never neither — check for that before constructing a request.",
-    "Self-serve accounts start with free runs per endpoint (2 each for text-to-music, text-to-sfx and audio-ducking; 1 each for the video endpoints), after which calls bill at the normal rate — so a first call succeeding is not evidence that the account has billing set up."
+    "Catch AuthenticationError (401), PaymentRequiredError (402), RateLimitError (429) and TaskFailedError (status \"failed\") separately rather than one generic except — callers usually handle these differently. All extend SoniloError.",
+    "video / video_url accept exactly one of the two, never both and never neither — validate before constructing a request.",
+    "Self-serve accounts start with free runs per endpoint (2 each for text-to-music, text-to-sfx, audio-ducking; 1 each for the video endpoints), then bill normally. A first call succeeding is not evidence that billing is set up."
   ]
 }
diff --git a/sonilo-cli/tests/test_context7.py b/sonilo-cli/tests/test_context7.py
index a7968b7..86d81b9 100644
--- a/sonilo-cli/tests/test_context7.py
+++ b/sonilo-cli/tests/test_context7.py
@@ -10,3 +10,40 @@ def test_context7_json_is_valid_and_complete():
     assert data["$schema"] == "https://context7.com/schema/context7.json"
     assert "**/__pycache__/**" in data["excludeFolders"]
     assert isinstance(data["rules"], list) and len(data["rules"]) >= 5
+
+
+# Limits from https://context7.com/schema/context7.json. The parser tolerates
+# oversized values, so nothing complains at index time — but the ownership
+# claim validates strictly and rejects the whole file, which is how five
+# over-long rules went unnoticed until claiming failed.
+MAX_RULE_LEN = 255
+MAX_RULES = 50
+MAX_DESCRIPTION_LEN = 200
+MAX_PROJECT_TITLE_LEN = 100
+
+ALLOWED_KEYS = {
+    "$schema", "projectTitle", "description", "branch", "folders",
+    "excludeFolders", "excludeFiles", "rules", "disallow", "redirect",
+    "previousVersions", "url", "public_key",
+}
+
+
+def test_context7_json_respects_schema_limits():
+    data = json.loads((ROOT / "context7.json").read_text())
+
+    unknown = set(data) - ALLOWED_KEYS
+    assert not unknown, f"unknown top-level keys reject the claim: {unknown}"
+
+    assert len(data["rules"]) <= MAX_RULES
+    too_long = {i: len(r) for i, r in enumerate(data["rules"]) if len(r) > MAX_RULE_LEN}
+    assert not too_long, f"rules over {MAX_RULE_LEN} chars: {too_long}"
+
+    assert len(data["description"]) <= MAX_DESCRIPTION_LEN
+    assert len(data["projectTitle"]) <= MAX_PROJECT_TITLE_LEN
+
+
+def test_context7_json_carries_ownership_keys():
+    """The claim flow requires both, and the schema makes them interdependent."""
+    data = json.loads((ROOT / "context7.json").read_text())
+    assert data["url"] == "https://context7.com/sonilo-ai/sonilo-python"
+    assert data["public_key"].startswith("pk_")