llm.rb 11.2.0 → 11.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bb1ffd1e0ecb17422014ec8f75c8b729f74d0a7cef4fd4e12681ef254411b24a
-  data.tar.gz: e5a7815d52c6fa99a38dec111c6d71aef9782272b97bfbad81e8f5bee913f918
+  metadata.gz: e3aad41c57654e1abef4c184da8904ae5840708e8f5ce937e1030279a5e5a240
+  data.tar.gz: 9b03ac6a4f04190a57f2e4698eab923087febbab6a1ff8ad67f391fce4bd5c1e
 SHA512:
-  metadata.gz: d7c2d1dac8ef97a5be2540896828b523ce1491e43f2fd8c78e53b8fbab34432bf6dedaee066afd43d261fb8f4e2fb9f7c3d8c112de83e60ee809d6ef77f41feb
-  data.tar.gz: d5073e8c5c156739cff4c68c65c4efaf88403229841ada3cdd1adb4920dfb58c00017ae7dbd73d12ef17b6c191b0b71ab4f15fd0c0ae29991dcffb1b840381d0
+  metadata.gz: f429a75d5b612eb27968b6e6451bd231f2191b5b955262503c68baa7ca27f289c01ad564558e82d7fcc732393f635c32caf883a36815b41d86f6c1289cfffb61
+  data.tar.gz: 90924b997b5d47d74c3df6e303c890fb5d1a771bae10e63c349436bcbd061241604184f62df87cb2dee7d85405fbd53f08d279ef6b4d3a85a12b0d59b7377229

data/CHANGELOG.md

@@ -2,6 +2,80 @@
 
 ## Unreleased
 
+## v11.3.1
+
+Changes since `v11.3.0`.
+
+This release rebrands the project under the r.uby.dev umbrella, removes
+the Jekyll-based docs site in favor of a pure-markdown deepdive, and
+cleans up YARD documentation across the codebase.
+
+### Change
+
+* **Rebrand to r.uby.dev** <br>
+  Update README.md with the new logo, streamlined copy, and r.uby.dev
+  URLs. Rewrite `resources/deepdive.md` as a concise walkthrough and
+  bundle it with the gem. Remove the `docs/` directory (Jekyll site).
+  Update all references from `llmrb.github.io` to `r.uby.dev`.
+
+* **Update gemspec** <br>
+  Update homepage, metadata URLs, email, and author list. Switch the
+  YARD markdown processor from kramdown to redcarpet.
+
+### Fix
+
+* **Fix YARD documentation** <br>
+  Fix unnamed, misnamed, and missing `@param` tags across provider
+  adapters, transport classes, stream, tool, schema, registry, agent,
+  and ActiveRecord integration files. Fix backtick-wrapped constant
+  references and other YARD formatting issues.
+
+## v11.3.0
+
+Changes since `v11.2.0`.
+
+This release promotes `LLM::Agent` as the default high-level runtime,
+raises `LLM::NotFoundError` for provider 404 responses, and adds
+Symbol resolution to `LLM::Agent.confirm` and `LLM::Agent.skills` for
+dynamic tool confirmation and skill lists.
+
+### Add
+
+* **Raise `LLM::NotFoundError` for provider 404 responses** <br>
+  Raise `LLM::NotFoundError` when a provider returns HTTP 404. One
+  example is calling the embeddings API on DeepSeek
+  (`LLM.deepseek(...).embed(["foobar"])`), which returns 404 because
+  DeepSeek does not implement that endpoint.
+
+* **Add Symbol resolution to `LLM::Agent.confirm`** <br>
+  When `confirm` receives a single Symbol argument, it stores it
+  as-is instead of converting it to a string array. At initialization
+  time, `resolve_option` resolves the Symbol by calling the method
+  with that name on the agent instance, and the result is converted
+  to strings. This allows dynamic tool confirmation lists:
+
+      class MyAgent < LLM::Agent
+        confirm :tools_that_need_confirmation
+
+        def tools_that_need_confirmation
+          some_condition ? %w[delete destroy] : %w[delete]
+        end
+      end
+
+  Ported from llmrb/mruby-llm@89a232e3 and @2dd04e2d.
+
+  Extend the same pattern to `LLM::Agent.skills` so the skills DSL
+  accepts a Symbol that resolves through the agent instance at
+  initialization time.
+
+### Change
+
+* **Clarify `LLM::Agent` as the default high-level runtime** <br>
+  Document that `LLM::Context` remains at the heart of llm.rb, but
+  `LLM::Agent` is the better default unless an application needs advanced
+  manual tool loops. `LLM::Agent` manages the tool loop for callers and
+  enables guards against runaway or repeated tool-call loops.
+
 ## v11.2.0
 
 Changes since `v11.1.0`.
@@ -222,7 +296,7 @@ requests outside `#session`, `LLM::Function#def` as a short alias for
   Fix block-form `model { ... }`, `tools { ... }`, and
   `schema { ... }` declarations in the ActiveRecord and Sequel agent
   wrappers so persisted agent models configure the internal agent class
-  the same way as `LLM::Agent`.
+  the same way `LLM::Agent` does.
 
 * **Fix missing `skills` in ORM agent wrappers** <br>
   Fix the ActiveRecord and Sequel agent wrappers to expose `skills`, so
@@ -465,7 +539,7 @@ DSML tool-marker filtering in streamed text.
   blocks that Bedrock rejects.
 
 * **Suppress Bedrock DSML tool markers in streamed text** <br>
-  Filter `\"<｜DSML｜function_calls\"` markers out of streamed Bedrock
+  Filter `"\u003c\u003cDSML\u003efunction_calls\u003e\u003e"` markers out of streamed Bedrock
   assistant text so tool-call sentinels do not leak into user-visible
   output.
 
@@ -475,7 +549,7 @@ Changes since `v7.0.0`.
 
 This release adds Unix-fork concurrency for process-isolated tool
 execution, extends `LLM::Object` with `#merge` and `#delete`, and drops
-Ruby 3.2 support due to segfaults observed with the `:fork` path. It
+Ruby 3.2 support due to a segfault observed with the `:fork` path. It
 promotes `LLM::Pipe` to the top-level namespace and adds
 `persistent: true` on `LLM::MCP.http` for direct persistent transport
 configuration. `LLM::Function#runner` is exposed as public API, agent
@@ -616,7 +690,7 @@ provider usage has been recorded yet.
   buffer API.
 
 * **Support percentage compaction token thresholds** <br>
-  Let `LLM::Compactor` accept `token_threshold:` values like `\"90%\"` so
+  Let `LLM::Compactor` accept `token_threshold:` values like `"90%"` so
   compaction can trigger at a percentage of the active model context
   window.
 
@@ -775,7 +849,7 @@ interruption use the active per-call stream correctly.
 
 * **Refresh provider model metadata** <br>
   Add current DeepSeek and OpenAI model metadata to `data/` and update the
-  Google Gemma model entry to match the current provider naming.
+  Google Gemini model entry to match the current provider naming.
 
 ### Fix
 
@@ -1216,12 +1290,12 @@ Changes since `v4.14.0`.
   storage when Sequel JSON typecasting is enabled.
 
 * **Improve streaming parser performance** <br>
-  In the local replay-based `stream_parser` benchmark versus
-  `v4.14.0` (median of 20 samples, 5000 iterations), plain Ruby is a
+  In the local replay-based `stream_parser` benchmark versus `v4.14.0`
+  (median of 20 samples, 5000 iterations), plain Ruby is a
   small overall win: the generic eventstream path is about 0.4%
   faster, the OpenAI stream parser is about 0.5% faster, and the
   OpenAI Responses parser is about 1.6% faster, with unchanged
-  allocations. Under YJIT on the same benchmark, the generic
+  allocations. Under YJIT on the same benchmark harness, the generic
   eventstream path is about 0.9% faster and the OpenAI stream parser
   is about 0.4% faster, while the OpenAI Responses parser is about
   0.7% slower, also with unchanged allocations.
@@ -1263,7 +1337,7 @@ parallel tool calls can safely share one connection.
 * **Reduce provider streaming allocations** <br>
   Decode streamed provider payloads directly in
   `LLM::Provider::Transport::HTTP` before handing them to provider
-  parsers, which cuts allocation churn and gives a smaller streaming
+  parsers, which cuts allocation churn and gives a small streaming
   speed bump.
 
 * **Reduce generic SSE parser allocations** <br>
@@ -1399,7 +1473,7 @@ Changes since `v4.9.0`.
 
 - Add HTTP transport for MCP with `LLM::MCP::Transport::HTTP` for remote servers
 - Add JSON Schema union types (`any_of`, `all_of`, `one_of`) with parser integration
-- Add JSON Schema type array union support (e.g., `\"type\": [\"object\", \"null\"]`)
+- Add JSON Schema type array union support (e.g., `"type": ["object", "null"]`)
 - Add JSON Schema type inference from `const`, `enum`, or `default` fields
 
 ### Change
@@ -1500,7 +1574,7 @@ Notable merged work in this range includes:
 - `Add rack + websocket example (#130)`
 - `feat(gemspec): add changelog URI (#136)`
 - `feat(function): alias ThreadGroup#wait as ThreadGroup#value (#62)`
-- README and screencast refresh across `#66`, `#68`, `#71`, and
+- `README and screencast refresh across `#66`, `#68`, `#71`, and
   `#72`
 - `chore(bot): update deprecation warning from v5.0 to v6.0`
 - `fix(deepseek): tolerate malformed tool arguments`