riffer 0.55.0 → 0.56.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (57) hide show
  1. checksums.yaml +4 -4
  2. data/.release-please-manifest.json +1 -1
  3. data/CHANGELOG.md +18 -0
  4. data/README.md +2 -1
  5. data/UPGRADING.md +3 -0
  6. data/docs/AGENT_LIFECYCLE.md +1 -1
  7. data/docs/CONFIGURATION.md +48 -1
  8. data/docs/GUARDRAILS.md +2 -0
  9. data/docs/MESSAGES.md +112 -15
  10. data/docs/STABILITY.md +32 -0
  11. data/docs/STREAM_EVENTS.md +16 -0
  12. data/docs/TESTING.md +22 -1
  13. data/docs/providers/AMAZON_BEDROCK.md +1 -1
  14. data/docs/providers/ANTHROPIC.md +18 -14
  15. data/docs/providers/CUSTOM_PROVIDERS.md +24 -0
  16. data/docs/providers/GEMINI.md +1 -1
  17. data/docs/providers/GOOGLE_CLOUD.md +1 -1
  18. data/docs/providers/MOCK_PROVIDER.md +2 -0
  19. data/docs/providers/OPENAI.md +3 -1
  20. data/docs-site/manifest.yml +3 -0
  21. data/lib/riffer/agent/run.rb +28 -13
  22. data/lib/riffer/agent/session.rb +33 -10
  23. data/lib/riffer/config/future.rb +53 -0
  24. data/lib/riffer/config/pricing.rb +5 -15
  25. data/lib/riffer/config.rb +16 -0
  26. data/lib/riffer/deprecation/event.rb +25 -0
  27. data/lib/riffer/deprecation.rb +60 -0
  28. data/lib/riffer/messages/assistant/server_tool_part.rb +73 -0
  29. data/lib/riffer/messages/assistant/text_part.rb +52 -0
  30. data/lib/riffer/messages/assistant.rb +64 -17
  31. data/lib/riffer/providers/anthropic_compatible/messages_api/stream.rb +25 -1
  32. data/lib/riffer/providers/anthropic_compatible/messages_api.rb +51 -29
  33. data/lib/riffer/providers/base.rb +15 -6
  34. data/lib/riffer/providers/open_ai.rb +21 -16
  35. data/lib/riffer/stream_events/server_tool_done.rb +21 -0
  36. data/lib/riffer/testing.rb +6 -0
  37. data/lib/riffer/version.rb +1 -1
  38. data/lib/riffer.rb +2 -0
  39. data/sig/generated/riffer/agent/run.rbs +5 -0
  40. data/sig/generated/riffer/agent/session.rbs +6 -0
  41. data/sig/generated/riffer/config/future.rbs +29 -0
  42. data/sig/generated/riffer/config/pricing.rbs +0 -9
  43. data/sig/generated/riffer/config.rbs +8 -0
  44. data/sig/generated/riffer/deprecation/event.rbs +17 -0
  45. data/sig/generated/riffer/deprecation.rbs +27 -0
  46. data/sig/generated/riffer/messages/assistant/server_tool_part.rbs +44 -0
  47. data/sig/generated/riffer/messages/assistant/text_part.rbs +35 -0
  48. data/sig/generated/riffer/messages/assistant.rbs +24 -11
  49. data/sig/generated/riffer/providers/anthropic_compatible/messages_api/stream.rbs +10 -0
  50. data/sig/generated/riffer/providers/anthropic_compatible/messages_api.rbs +18 -9
  51. data/sig/generated/riffer/providers/base.rbs +5 -0
  52. data/sig/generated/riffer/providers/open_ai.rbs +7 -2
  53. data/sig/generated/riffer/stream_events/server_tool_done.rbs +15 -0
  54. data/sig/generated/riffer/testing.rbs +4 -0
  55. data/sig/generated/riffer.rbs +3 -0
  56. data/sig/manual/riffer/deprecation.rbs +5 -0
  57. metadata +16 -1
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: bce35dc408dca35e5effe8987b7a8a43865720704f4c3c5f47e1c7709c733bb4
4
- data.tar.gz: 0edac840e1a65f7cf55ef5254ed6451b2c0aba469fed02fcb62eb4891970236f
3
+ metadata.gz: 4eb9fc9eb011b6c16fe2e87bb6dcfc3a8a5ba3f0ae3b29b72f530e61fe1d300d
4
+ data.tar.gz: 9016cfeb1bde20ce4788673b1d6398c1f087ec47893c575cf8fc4998fd797b31
5
5
  SHA512:
6
- metadata.gz: b23684154e93a57612c4de3b860336804ae3ccf265a951bcc71126bcc71d8f49de1e42331418187243ce846b88c47a4dcc20d52a196680f023c5ebb9da049b9d
7
- data.tar.gz: daaf1a51e35cd78605b82ddd11608713a0c1c2d231d8a7eae30d4048f0d60793e2c9a99aa5bbc3af6c48dd52f36ac095c860cdb7a665f39de319f6066182a1c0
6
+ metadata.gz: c9603251a779743eb3b8a8ad5d2b289924bd1e678595ec714b89602f5bea1c515516f9f7a2521ce0f67511bd602db0a2cb83d47c45b9cf364aead3aa0039c1b0
7
+ data.tar.gz: c243d5ca21b60801779c002bca9badc513ce6179066aa223f1351763f1aaded7d02fd37ac2695649aeab76ac7ed4a0d448e7f95feaf3a8b8aae722cd622f11d7
@@ -1,3 +1,3 @@
1
1
  {
2
- ".": "0.55.0"
2
+ ".": "0.56.0"
3
3
  }
data/CHANGELOG.md CHANGED
@@ -5,6 +5,24 @@ All notable changes to this project will be documented in this file.
5
5
  The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
6
6
  and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
7
7
 
8
+ ## [0.56.0](https://github.com/janeapp/riffer/compare/riffer/v0.55.0...riffer/v0.56.0) (2026-11-09)
9
+
10
+
11
+ ### ⚠ BREAKING CHANGES
12
+
13
+ * **messages:** `Assistant#to_h` writes `parts` instead of `content`, `tool_calls` and `reasoning`, and `to_h` no longer accepts `parts:`. Build the old fields from `#content`, `#tool_calls` and `#reasoning` on the message. This also changes the assistant entries in `Riffer::Providers::Mock#calls` and `Riffer::Evals::ScenarioResult#to_h`.
14
+
15
+ ### Features
16
+
17
+ * add Riffer::Deprecation for deprecation warnings ([#549](https://github.com/janeapp/riffer/issues/549)) ([6ad6046](https://github.com/janeapp/riffer/commit/6ad604647de1074d4f3f0739727940bcf06667fd))
18
+ * add Riffer.config.future for opt-in future flags ([#556](https://github.com/janeapp/riffer/issues/556)) ([6e49471](https://github.com/janeapp/riffer/commit/6e4947197d01672466606cd10541e39a2dcd521a))
19
+ * **messages:** keep assistant turns in order with Assistant#parts ([#527](https://github.com/janeapp/riffer/issues/527)) ([15e8eb9](https://github.com/janeapp/riffer/commit/15e8eb935f588a3937df449449d38712b0d12414))
20
+
21
+
22
+ ### Bug Fixes
23
+
24
+ * **openai:** return web search sources and drop tool calls cut off by the token limit ([#541](https://github.com/janeapp/riffer/issues/541)) ([659e136](https://github.com/janeapp/riffer/commit/659e13672031f1867a5fd5345468db20487d3f27))
25
+
8
26
  ## [0.55.0](https://github.com/janeapp/riffer/compare/riffer/v0.54.0...riffer/v0.55.0) (2026-10-29)
9
27
 
10
28
 
data/README.md CHANGED
@@ -63,6 +63,7 @@ Comprehensive guides live at [riffer.ai](https://riffer.ai):
63
63
  - [MCP](https://riffer.ai/guides/mcp/) - Integrating third-party MCP servers
64
64
  - [Serialization](https://riffer.ai/guides/serialization/) - Persisting and transferring agent definitions
65
65
  - [Tracing](https://riffer.ai/guides/tracing/) - OpenTelemetry span contract and host wiring
66
+ - [Stability](https://riffer.ai/guides/stability/) - What's safe to depend on and how upgrades work
66
67
  - [Providers](https://riffer.ai/guides/providers/overview/) - LLM provider adapters
67
68
 
68
69
  The guide sources are in the [docs](docs/) directory.
@@ -159,7 +160,7 @@ Please follow the [Code of Conduct](https://github.com/janeapp/riffer/blob/main/
159
160
 
160
161
  ## Changelog
161
162
 
162
- All notable changes to this project are documented in `CHANGELOG.md`.
163
+ All notable changes to this project are documented in `CHANGELOG.md`. Breaking releases come with an upgrade guide in [UPGRADING.md](UPGRADING.md).
163
164
 
164
165
  ## License
165
166
 
data/UPGRADING.md ADDED
@@ -0,0 +1,3 @@
1
+ # Upgrading
2
+
3
+ Upgrade guides for breaking releases will appear here. See the [stability policy](https://riffer.ai/guides/stability/) for when breaking changes can happen. To try a breaking release's new behaviour early, turn on its [future flag](https://riffer.ai/guides/configuration/#future-flags).
@@ -270,7 +270,7 @@ If you need finer control over placeholder content (per-call shape, structured m
270
270
 
271
271
  The session exposes a small set of in-place mutators that enforce the `tool_use` ↔ `tool_result` invariant on every operation. Use these to align history with external state (persisted transcript, partial output that wasn't actually delivered, etc.) without rebuilding the agent.
272
272
 
273
- - **`agent.session.update(id:, **attrs)`** — In-place partial update. Looks up by message `id:`; builds a replacement of the same type with `attrs` overlaid on the existing fields. Use this to edit assistant content (`update(id:, content:)`), restate a system message, etc. When the target is an assistant and the update drops entries from `tool_calls`, matching `Tool` children are removed atomically.
273
+ - **`agent.session.update(id:, **attrs)`** — In-place partial update. Looks up by message `id:`; builds a replacement of the same type with `attrs` overlaid on the existing fields. Use this to edit assistant content (`update(id:, content:)`), restate a system message, etc. When the target is an assistant and the update drops entries from `tool_calls`, matching `Tool` children are removed atomically. An assistant update with `content:`, `tool_calls:` or `reasoning:` rebuilds the turn's [parts](MESSAGES.md#parts) in the default order, which drops server tool parts; pass `parts:` to replace them as given.
274
274
  - **`agent.session.update(tool_call_id:, **attrs)`** — Same as above but looks up the tool result by `tool_call_id:`. Preserves `name` and `id`. Use this to upgrade a `discard_pending_tool_calls` placeholder once the real result is available (`update(tool_call_id:, content:, error: nil, error_type: nil)`).
275
275
  - **`agent.session.remove(id:)`** — Removes a message; cascades to its `Tool` children when the target carries `tool_calls`. Raises if called on a `Tool` message (use `update(tool_call_id:, ...)` to rewrite a tool result instead).
276
276
  - **`agent.session.discard_pending_tool_calls`** — Answers every unanswered `tool_use` with an `:interrupted` placeholder result and returns the filled `call_id`s (see "Discarding pending tool calls after an interrupt" above).
@@ -276,7 +276,7 @@ See [Model Catalog](CATALOG.md) for the file format, lookup, layering, and valid
276
276
 
277
277
  ### Pricing (deprecated)
278
278
 
279
- `config.pricing` is deprecated and will be removed in a future release. Move rates into a [catalog file](CATALOG.md#pricing); the first `set` call prints a deprecation warning. Riffer still reads these rates as a fallback, and catalog pricing wins when both price a model.
279
+ `config.pricing` is deprecated and will be removed in a future release. Move rates into a [catalog file](CATALOG.md#pricing); each line that calls `set` reports a [deprecation](#deprecations). Riffer still reads these rates as a fallback, and catalog pricing wins when both price a model.
280
280
 
281
281
  ```ruby
282
282
  # before
@@ -291,6 +291,53 @@ end
291
291
 
292
292
  `set` takes a `provider/model` id or an array of ids, plus `input:`, `output:`, and optional `cache_read:` and `cache_write:` rates per million tokens. The rates and cost formula are the same as [catalog pricing](CATALOG.md#pricing).
293
293
 
294
+ ### Deprecations
295
+
296
+ A deprecated riffer call reports itself at least one release before it is removed. The report names the line in your code that made the call:
297
+
298
+ ```
299
+ riffer: config.pricing is deprecated; move rates into a catalog file listed in config.catalog_files (see docs/CATALOG.md). Called from config/initializers/riffer.rb:4.
300
+ ```
301
+
302
+ `config.deprecations` decides what happens:
303
+
304
+ ```ruby
305
+ Riffer.configure do |config|
306
+ config.deprecations = ->(event) { Rails.logger.warn(event.to_s) }
307
+ end
308
+ ```
309
+
310
+ | Value | Description |
311
+ | ----------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
312
+ | `:warn` (default) | Prints the report to stderr with `Kernel.warn`, once per deprecation per calling line. |
313
+ | `:raise` | Raises `Riffer::DeprecationError` on every call, so a test suite fails on each test that makes one. See [Testing — Deprecations](TESTING.md#deprecations). |
314
+ | `:silence` | Reports nothing. |
315
+ | A callable | Called with a `Riffer::Deprecation::Event` once per deprecation per calling line. The event has `id`, `message` and `location` (`"path:line"`); `to_s` is the report. |
316
+
317
+ Any other value raises `Riffer::ArgumentError`.
318
+
319
+ ### Future flags
320
+
321
+ `config.future` lets you turn on new behaviour before it becomes the default, so you can try it and adapt ahead of the breaking release that switches it on for everyone:
322
+
323
+ ```ruby
324
+ Riffer.configure do |config|
325
+ config.future.some_flag = true
326
+ end
327
+ ```
328
+
329
+ A flag accepts booleans, `'true'`/`'false'`/`'1'`/`'0'`, `1`/`0`, or `nil` (which turns an opt-in flag off); any other value raises `Riffer::ArgumentError`. Setting a flag Riffer doesn't know raises `NoMethodError`.
330
+
331
+ A flag goes through three stages:
332
+
333
+ 1. **Opt-in.** The flag is off by default, and setting it to `true` turns the new behaviour on.
334
+ 2. **Default.** A [breaking release](STABILITY.md#breaking-releases) makes the new behaviour the default. The flag always reads as on, and setting it reports a [deprecation](#deprecations) saying it's no longer needed and can be removed.
335
+ 3. **Removed.** A release later the flag is gone, and setting it raises `NoMethodError`.
336
+
337
+ | Flag | Stage | Description |
338
+ | -------- | ----- | ------------------------------- |
339
+ | _(none)_ | | Riffer has no future flags yet. |
340
+
294
341
  ### Message ID Strategy
295
342
 
296
343
  Opt in to stable identifiers on every message for logging, persistence, or replay:
data/docs/GUARDRAILS.md CHANGED
@@ -333,6 +333,8 @@ class TokenLimiter < Riffer::Guardrail
333
333
  end
334
334
  ```
335
335
 
336
+ `Assistant.new(text)` builds a new turn with only that text, so this example also drops the response's tool calls and server tool parts. When a guardrail changes the text of a response that carries reasoning, server tool parts or tool calls, rebuild its `parts` instead; see [Messages — Editing a turn](MESSAGES.md#parts).
337
+
336
338
  ### Content Policy Filter
337
339
 
338
340
  ```ruby
data/docs/MESSAGES.md CHANGED
@@ -57,7 +57,7 @@ msg = Riffer::Messages::Assistant.new("", tool_calls: [
57
57
  {id: "call_123", call_id: "call_123", name: "weather_tool", arguments: '{"city":"Tokyo"}'}
58
58
  ])
59
59
  msg.tool_calls # => [{id: "call_123", ...}]
60
- msg.to_h # => {role: "assistant", content: "", tool_calls: [...]}
60
+ msg.to_h # => {role: :assistant, parts: [{part: :tool_call, ...}]}
61
61
 
62
62
  # Accessing token usage data (when available from provider)
63
63
  if msg.token_usage
@@ -134,7 +134,7 @@ The `to_h` representation includes `structured_output` only when present:
134
134
 
135
135
  ```ruby
136
136
  msg = Riffer::Messages::Assistant.new('{"sentiment":"positive"}', structured_output: {sentiment: "positive"})
137
- msg.to_h # => {role: :assistant, content: '{"sentiment":"positive"}', structured_output: {sentiment: "positive"}}
137
+ msg.to_h # => {role: :assistant, parts: [{part: :text, text: '{"sentiment":"positive"}'}], structured_output: {sentiment: "positive"}}
138
138
  ```
139
139
 
140
140
  #### Reasoning
@@ -170,13 +170,14 @@ A `type` outside the three values raises `Riffer::ArgumentError`. `format` is a
170
170
 
171
171
  `reasoning?` is true when the message carries any part. `reasoning_text` joins the `text` of the `:text` and `:summary` parts with blank lines, skipping `:encrypted` parts, and is `nil` when there is nothing to join. The run's final assistant message projects its parts onto `response.reasoning` (see [Agent Lifecycle — Response Attributes](AGENT_LIFECYCLE.md#response-attributes)).
172
172
 
173
- Parts round-trip through `to_h` / `from_hash` like every other message field, so an application that persists sessions can store and replay them. The `reasoning` key is absent from `to_h` when the message has no parts, and each part omits the fields it doesn't carry:
173
+ Parts round-trip through `to_h` / `from_hash` like every other message field, so an application that persists sessions can store and replay them. `to_h` writes each reasoning part as a [`parts`](#parts) entry tagged `part: :reasoning`, and each part omits the fields it doesn't carry:
174
174
 
175
175
  ```ruby
176
176
  msg.to_h
177
- # => {role: :assistant, content: "42", reasoning: [
178
- # {type: :summary, text: "The user wants the answer.", format: "mock-v1"},
179
- # {type: :encrypted, data: "b3BhcXVl", format: "mock-v1"}
177
+ # => {role: :assistant, parts: [
178
+ # {part: :reasoning, type: :summary, text: "The user wants the answer.", format: "mock-v1"},
179
+ # {part: :reasoning, type: :encrypted, data: "b3BhcXVl", format: "mock-v1"},
180
+ # {part: :text, text: "42"}
180
181
  # ]}
181
182
 
182
183
  Riffer::Messages::Base.from_hash(msg.to_h).reasoning # => [ReasoningPart, ReasoningPart]
@@ -184,14 +185,108 @@ Riffer::Messages::Base.from_hash(msg.to_h).reasoning # => [ReasoningPart, Reaso
184
185
 
185
186
  **The replay contract.** A provider adapter replays only the parts whose `format` it recognizes and silently skips the rest, so history that travelled through another provider is never rejected. A part with no `format` is never replayed; adapters that surface reasoning text but cannot yet send it back emit their parts that way, so the text is kept for display without risking a rejected request. Parts are never reordered, merged, or edited — riffer treats them as opaque, because the provider's signature covers their exact bytes.
186
187
 
187
- An application that would rather not store parts can leave the key out when it serializes:
188
+ An application that would rather not store reasoning can drop the reasoning entries of `parts` when it serializes:
188
189
 
189
190
  ```ruby
190
- msg.to_h.except(:reasoning)
191
+ hash = msg.to_h
192
+ hash[:parts] = hash[:parts].reject { |part| part[:part] == :reasoning }
191
193
  ```
192
194
 
193
195
  To drop parts from the in-memory session mid-run instead, `Session#update(id:, reasoning: [])` rewrites the message in place; it needs [message ids](#ids) enabled to address it.
194
196
 
197
+ #### Parts
198
+
199
+ `parts` is the assistant turn in the order the provider produced it. `content`, `tool_calls` and `reasoning` are views over it. Some providers interleave reasoning, server-side tools and text in one turn, and Anthropic needs the turn back in that order: Claude signs each thinking block over every block before it. Sending earlier turns back unchanged also keeps the model's own reasoning and search results, and keeps the prompt cache warm, since the history doesn't change between requests.
200
+
201
+ A part is one of:
202
+
203
+ | Class | Holds |
204
+ | ---------------- | -------------------------------------------------------------------------------------------------------------------------------------- |
205
+ | `TextPart` | One stretch of answer text (`text`). `data` keeps the provider's block when it carries more than the text, such as Anthropic citations |
206
+ | `ReasoningPart` | A thinking block (see [Reasoning](#reasoning)) |
207
+ | `ToolCall` | A function call riffer runs (see [Tool Calls](#tool-calls)) |
208
+ | `ServerToolPart` | A call (`kind: :call`) or result (`kind: :result`) of a tool the provider ran itself, such as web search |
209
+
210
+ `TextPart` carries:
211
+
212
+ | Field | Type | Description |
213
+ | -------- | --------- | ------------------------------------------------------------------------------------------------------ |
214
+ | `text` | `String` | The answer text |
215
+ | `data` | `String?` | The provider's whole block as JSON, kept when it carries more than the text (e.g. Anthropic citations) |
216
+ | `format` | `String?` | The wire format of `data`, owned by the adapter that produced it |
217
+
218
+ `ServerToolPart` carries:
219
+
220
+ | Field | Type | Description |
221
+ | -------- | --------- | --------------------------------------------------------------------------------------------- |
222
+ | `name` | `String` | The tool's name (e.g. `"web_search"`) |
223
+ | `kind` | `Symbol` | `:call` or `:result` |
224
+ | `data` | `String` | The provider's block as JSON, unchanged |
225
+ | `id` | `String?` | The provider's identifier, shared by a call and its result |
226
+ | `format` | `String?` | The wire format, owned by the adapter that produced the part (e.g. `"anthropic-messages-v1"`) |
227
+
228
+ A `kind` outside `:call` and `:result` raises `Riffer::ArgumentError`.
229
+
230
+ ```ruby
231
+ msg = Riffer::Messages::Assistant.new(parts: [
232
+ Riffer::Messages::Assistant::TextPart.new("Let me check. "),
233
+ Riffer::Messages::Assistant::ServerToolPart.new(name: "web_search", kind: :call, id: "srvtoolu_1", data: "{...}", format: "anthropic-messages-v1"),
234
+ Riffer::Messages::Assistant::ServerToolPart.new(name: "web_search", kind: :result, id: "srvtoolu_1", data: "{...}", format: "anthropic-messages-v1"),
235
+ Riffer::Messages::Assistant::ReasoningPart.new(type: :text, text: "", signature: "CAQS...", format: "anthropic-messages-v1"),
236
+ Riffer::Messages::Assistant::TextPart.new("The latest stable Ruby version is 4.0.9.")
237
+ ])
238
+
239
+ msg.content # => "Let me check. The latest stable Ruby version is 4.0.9."
240
+ msg.reasoning # => [ReasoningPart]
241
+ msg.reasoning? # => true; server tool parts don't count as reasoning
242
+ ```
243
+
244
+ - `content` joins the text parts with no separator.
245
+ - `Assistant.new(content, tool_calls:, reasoning:)` builds the parts in a fixed order: reasoning, then `content` as one text part (none when empty), then tool calls.
246
+ - Passing `parts:` together with `content`, `tool_calls:` or `reasoning:` raises `Riffer::ArgumentError`.
247
+ - `ServerToolPart#data` holds the provider's block as JSON, unchanged. It follows the same replay contract as reasoning: only the adapter that recognizes its `format` sends it back, and other adapters skip it. The same goes for a `TextPart` that carries `data`; other adapters send its `text`.
248
+
249
+ `to_h` writes the turn as a `parts` key, so it keeps the order, the server tool parts and the `data` of text parts. Each entry is the part's own hash plus a `part` tag (`:text`, `:reasoning`, `:tool_call` or `:server_tool`). It doesn't write `content`, `tool_calls` or `reasoning`; read those from the message:
250
+
251
+ ```ruby
252
+ msg.to_h
253
+ # => {role: :assistant, parts: [
254
+ # {part: :text, text: "Let me check. "},
255
+ # {part: :server_tool, name: "web_search", kind: :call, data: "{...}", id: "srvtoolu_1", format: "anthropic-messages-v1"},
256
+ # ...
257
+ # ]}
258
+
259
+ {content: msg.content, tool_calls: msg.tool_calls.map(&:to_h), reasoning: msg.reasoning.map(&:to_h)}
260
+ ```
261
+
262
+ `from_hash` reads this shape and the older one with `content`, `tool_calls` and `reasoning` keys, so stored sessions still load. Rebuilt from the older shape, the turn is in the fixed order, without server tool parts or the `data` of text parts. A hash with `parts` and a non-nil `content`, or a non-empty `tool_calls` or `reasoning`, raises `Riffer::ArgumentError`.
263
+
264
+ **Editing a turn.** Riffer sends the parts back as they are. When you change a turn's text (in an output guardrail, or with `Session#update`), keeping the rest of the turn consistent is up to you:
265
+
266
+ - Drop that turn's reasoning parts. They were signed over the old text, so Anthropic would reject them or drop them silently, and they may still hold the text you removed.
267
+ - Keep server tool parts and tool calls, so the model doesn't search again and tool results keep their call.
268
+ - Put the new text after the last server tool part and before the first tool call.
269
+
270
+ ```ruby
271
+ def redact(message, text)
272
+ kept = message.parts.reject do |part|
273
+ part.is_a?(Riffer::Messages::Assistant::ReasoningPart) || part.is_a?(Riffer::Messages::Assistant::TextPart)
274
+ end
275
+ index = kept.index { |part| part.is_a?(Riffer::Messages::Assistant::ToolCall) } || kept.length
276
+ parts = kept.insert(index, Riffer::Messages::Assistant::TextPart.new(text))
277
+
278
+ Riffer::Messages::Assistant.new(
279
+ parts: parts,
280
+ id: message.id,
281
+ token_usage: message.token_usage,
282
+ finish_reason: message.finish_reason,
283
+ finish_reason_raw: message.finish_reason_raw
284
+ )
285
+ end
286
+ ```
287
+
288
+ `Assistant.new(text)` and `Session#update(id:, content:)` don't follow these rules. They rebuild the turn in the fixed order, which keeps the reasoning and drops server tool parts. Use `Session#update(id:, parts:)` to replace the parts as given.
289
+
195
290
  ### Tool
196
291
 
197
292
  Tool messages contain the results of tool executions:
@@ -360,12 +455,14 @@ Without this step, the same model can receive different input depending on the p
360
455
 
361
456
  ### Merge rules
362
457
 
363
- | Message type | Content | Auxiliary data | Merged? |
364
- | ------------ | -------------------- | -------------------------------- | ---------------------------------------- |
365
- | `System` | Joined with `"\n\n"` | — | Yes |
366
- | `User` | Joined with `"\n\n"` | `files` arrays concatenated | Yes |
367
- | `Assistant` | Joined with `"\n\n"` | `tool_calls` arrays concatenated | Yes |
368
- | `Tool` | — | — | Never (each has a unique `tool_call_id`) |
458
+ | Message type | Content | Auxiliary data | Merged? |
459
+ | ------------ | -------------------- | ------------------------------------------------ | ---------------------------------------- |
460
+ | `System` | Joined with `"\n\n"` | — | Yes |
461
+ | `User` | Joined with `"\n\n"` | `files` arrays concatenated | Yes |
462
+ | `Assistant` | Joined with `"\n\n"` | `tool_calls` and `reasoning` arrays concatenated | Yes |
463
+ | `Tool` | — | — | Never (each has a unique `tool_call_id`) |
464
+
465
+ When either assistant message has [parts](#parts) its fields can't rebuild, the merged message concatenates the two `parts` lists instead, with a `"\n\n"` text part between the texts.
369
466
 
370
467
  ### Example
371
468
 
@@ -434,7 +531,7 @@ class Riffer::Messages::Base
434
531
  end
435
532
  ```
436
533
 
437
- Subclasses implement `role` and optionally extend `to_h` with additional fields.
534
+ Subclasses implement `role` and optionally extend `to_h` with additional fields. `Assistant` writes `parts` in place of `content`.
438
535
 
439
536
  ## Editing history after the fact
440
537
 
data/docs/STABILITY.md ADDED
@@ -0,0 +1,32 @@
1
+ # Stability
2
+
3
+ This page covers which parts of Riffer you can depend on, how changes to them are announced, and how breaking releases work. Riffer is not at 1.0 yet; the [roadmap](https://github.com/janeapp/riffer/issues/528) tracks the path there.
4
+
5
+ ## Public API
6
+
7
+ The public API is everything documented in these guides. You can build on it, and it changes only in the ways this page describes.
8
+
9
+ Everything else is internal, even when Ruby lets you reach it. That includes classes, methods and constants that appear in the [API reference](https://riffer.ai/api/) but not in a guide. Internal code can change in any release without notice. At 1.0, every public constant will be documented and everything else marked private.
10
+
11
+ ## Experimental features
12
+
13
+ Experimental features are labelled _experimental_ in the guides, like the [prompted structured output strategy](AGENTS.md#structured-output-strategy-experimental). They can change in any minor release, and the [CHANGELOG](https://github.com/janeapp/riffer/blob/main/CHANGELOG.md) notes each change.
14
+
15
+ ## Deprecations
16
+
17
+ A breaking change to the public API lands first as a new API or an [opt-in flag](CONFIGURATION.md#future-flags), and the old way prints a deprecation warning. The warning appears for at least one release before the old way is removed, and it names the earliest version it can be removed in. For larger changes, it also names a date.
18
+
19
+ ## Breaking releases
20
+
21
+ Before 1.0, any minor release can include breaking changes, but never without notice: an earlier release deprecates what's changing first, so you'll see the warnings before anything breaks. Each breaking release comes with an upgrade guide in [UPGRADING.md](https://github.com/janeapp/riffer/blob/main/UPGRADING.md), and the CHANGELOG lists its changes under **⚠ BREAKING CHANGES**.
22
+
23
+ After 1.0, breaking changes ship only in major versions.
24
+
25
+ ## Saved formats
26
+
27
+ Saved formats are the data Riffer produces for you to store and load later, such as [serialized agents](SERIALIZATION.md) and [messages persisted with `to_h`](MESSAGES.md#parts).
28
+
29
+ - Riffer can read a new format at least one release before it starts writing it, so a deploy that runs a release alongside the one before it keeps working.
30
+ - Anything saved by a 1.x release loads in every later 1.x release.
31
+
32
+ See [Serialization → Versioning](SERIALIZATION.md#versioning) for how agent data records its format version.
@@ -55,6 +55,8 @@ event.to_h # => {role: :assistant, content: "Hello, how can I help you?"}
55
55
 
56
56
  Fires once per model response, after its last `TextDelta`, and contains the response's full text: every text block the provider returned, concatenated in order, so it always equals the joined `TextDelta` contents. A response that carries no text (for example, only tool calls) emits no `TextDone`.
57
57
 
58
+ The agent loop builds the message's [text parts](MESSAGES.md#parts) from the `TextDelta` events: the deltas between two other parts (reasoning, a tool call, a server tool block) form one text part. An adapter that sends no deltas gets one text part, from `TextDone`, at the end of the turn.
59
+
58
60
  ### ToolCallDelta
59
61
 
60
62
  Emitted when tool call arguments are being streamed:
@@ -156,6 +158,20 @@ event.to_h # => {role: :assistant, query: "Ruby language", sources: [...]}
156
158
 
157
159
  Contains the search query and an array of source hashes with `title` and `url` keys.
158
160
 
161
+ ### ServerToolDone
162
+
163
+ Emitted when a block from a tool the provider ran itself (such as web search) is complete. It carries the block for replay; use `WebSearchStatus` and `WebSearchDone` to show progress.
164
+
165
+ ```ruby
166
+ part = Riffer::Messages::Assistant::ServerToolPart.new(name: "web_search", kind: :call, data: "{...}", format: "anthropic-messages-v1")
167
+ event = Riffer::StreamEvents::ServerToolDone.new(part)
168
+ event.role # => :assistant
169
+ event.part # => the ServerToolPart
170
+ event.to_h # => {role: :assistant, part: {name: "web_search", kind: :call, data: "{...}", format: "anthropic-messages-v1"}}
171
+ ```
172
+
173
+ The agent loop adds `part` to the assistant message's [parts](MESSAGES.md#parts). Only Anthropic emits it today.
174
+
159
175
  ### GuardrailTripwire
160
176
 
161
177
  Emitted when a guardrail blocks execution during streaming:
data/docs/TESTING.md CHANGED
@@ -16,7 +16,7 @@ require 'riffer/testing/rspec'
16
16
  require 'riffer/testing/minitest'
17
17
  ```
18
18
 
19
- Either one adds `stub_agent` and `stub_tool` to every example and cleans up after each one. Neither RSpec nor minitest is a riffer dependency; the adapter files are only loaded when you require them yourself.
19
+ Either one adds `stub_agent`, `stub_tool` and `allow_deprecations` to every example and cleans up after each one. Neither RSpec nor minitest is a riffer dependency; the adapter files are only loaded when you require them yourself.
20
20
 
21
21
  ## Stubbing a tool
22
22
 
@@ -110,6 +110,27 @@ end
110
110
 
111
111
  Tracking lives on `Riffer::Testing` itself, so `Riffer::Testing.stub_tool(...)` outside an example and `stub_tool(...)` inside one share one list. Tracking is not synchronized — stub from a single-threaded test, before concurrent lookups begin.
112
112
 
113
+ ## Deprecations
114
+
115
+ Make your suite fail when it hits a deprecated riffer call, so you find each one before an upgrade removes it:
116
+
117
+ ```ruby
118
+ # test/test_helper.rb
119
+ Riffer.config.deprecations = :raise
120
+ ```
121
+
122
+ Every deprecated call then raises `Riffer::DeprecationError`, naming the line that made it. A test that uses a deprecated call on purpose wraps it in `allow_deprecations`:
123
+
124
+ ```ruby
125
+ it 'still reads legacy pricing' do
126
+ allow_deprecations { Riffer.config.pricing.set('openai/gpt-5', input: 1.0, output: 2.0) }
127
+
128
+ # ...
129
+ end
130
+ ```
131
+
132
+ `allow_deprecations` silences deprecations until the block returns or raises, and returns the block's value. It applies to the current thread or fiber and to threads and fibers started inside the block; other threads keep reporting. It is also available as `Riffer::Testing.allow_deprecations` outside an example. See [Configuration — Deprecations](CONFIGURATION.md#deprecations) for the other modes.
133
+
113
134
  ## When a stub leaks
114
135
 
115
136
  Stubbing an identifier that is already taken raises `Riffer::DuplicateIdentifierError`:
@@ -195,7 +195,7 @@ The agent-level [`reasoning`](../AGENTS.md#reasoning) setting defaults to Claude
195
195
 
196
196
  Riffer sends the reasoning back to Bedrock on every later turn. How much of it Claude uses depends on the model; some only use the reasoning from the current tool-calling turn. For tool calls, sending it back is required: when thinking is enabled, Claude needs its earlier reasoning back to carry on after a tool result. You don't have to do anything; it happens as long as the assistant messages stay in the history.
197
197
 
198
- If you persist sessions, keep the `reasoning` key when you store messages (see [Messages — Reasoning](../MESSAGES.md#reasoning)). If you drop it, later turns lose the model's earlier reasoning, and a tool-calling turn with thinking enabled may be rejected.
198
+ If you persist sessions, keep the reasoning when you store messages (see [Messages — Reasoning](../MESSAGES.md#reasoning)). If you drop it, later turns lose the model's earlier reasoning, and a tool-calling turn with thinking enabled may be rejected.
199
199
 
200
200
  Only reasoning that Bedrock produced is sent back to Bedrock. A conversation that switches providers midway still works: reasoning from other providers is kept on the messages but left out of Bedrock requests.
201
201
 
@@ -29,12 +29,12 @@ Riffer.configure do |config|
29
29
  end
30
30
  ```
31
31
 
32
- | Option | Default | Description |
33
- | ------------ | -------------------------------------- | ------------------------------------------------------------------- |
34
- | `api_key` | required | Sent as `Authorization: Bearer <api_key>` |
35
- | `base_url` | `https://api.anthropic.com` | API origin; requests go to `<base_url>/v1/messages` |
36
- | `headers` | `{}` | Sent on every request; replaces riffer's headers of the same name |
37
- | `connection` | Faraday defaults, 10s open / 600s read | Your own `Faraday::Connection`; see [Connections][connections] |
32
+ | Option | Default | Description |
33
+ | ------------ | -------------------------------------- | ----------------------------------------------------------------- |
34
+ | `api_key` | required | Sent as `Authorization: Bearer <api_key>` |
35
+ | `base_url` | `https://api.anthropic.com` | API origin; requests go to `<base_url>/v1/messages` |
36
+ | `headers` | `{}` | Sent on every request; replaces riffer's headers of the same name |
37
+ | `connection` | Faraday defaults, 10s open / 600s read | Your own `Faraday::Connection`; see [Connections][connections] |
38
38
 
39
39
  Timeouts, retries, proxies, beta headers and errors are covered in [Configuration → Connections][connections]. An API error becomes a `Riffer::APIError` whose message carries the Anthropic error type, e.g. `Anthropic API error (401): authentication_error: Invalid bearer token`. No requests are retried unless your connection adds retries.
40
40
 
@@ -189,9 +189,11 @@ The reasoning is kept on the assistant message as [reasoning parts](../MESSAGES.
189
189
 
190
190
  ### Reasoning Replay
191
191
 
192
- Riffer sends the reasoning back to Anthropic on every later turn, as `thinking` and `redacted_thinking` blocks ahead of the message's text and `tool_use` blocks, in their original order and unchanged. For tool calls, sending it back is required: when thinking is enabled, Claude needs its earlier thinking blocks back to carry on after a tool result. You don't have to do anything; it happens as long as the assistant messages stay in the history.
192
+ Riffer sends every earlier turn back to Anthropic as the API returned it: the `thinking`, `redacted_thinking`, text, web search and `tool_use` blocks, in their original order and unchanged (see [Messages — Parts](../MESSAGES.md#parts)). Order matters because Claude signs each thinking block over everything before it, including earlier blocks of the same turn. A turn sent back in a different order loses that reasoning, or is rejected with a 400 on accounts that enforce the binding. For tool calls, sending the reasoning back is required: when thinking is enabled, Claude needs its earlier thinking blocks back to carry on after a tool result. You don't have to do anything; it happens as long as the assistant messages stay in the history.
193
193
 
194
- If you persist sessions, keep the `reasoning` key when you store messages (see [Messages — Reasoning](../MESSAGES.md#reasoning)). If you drop it, later turns lose the model's earlier reasoning, and a tool-calling turn with thinking enabled may be rejected.
194
+ If you persist sessions, `to_h` keeps the web search results and the turn's block order (see [Messages — Parts](../MESSAGES.md#parts)). Dropping the reasoning loses the model's earlier thinking, and a tool-calling turn with thinking enabled may be rejected.
195
+
196
+ If you change a turn's text, drop that turn's reasoning yourself; see [Editing a turn](../MESSAGES.md#parts).
195
197
 
196
198
  Only reasoning that this provider produced is sent back to Anthropic. A conversation that switches providers midway still works: reasoning from other providers (including OpenRouter's `anthropic-claude-v1` parts) is kept on the messages but left out of Anthropic requests.
197
199
 
@@ -222,16 +224,18 @@ end
222
224
 
223
225
  Anthropic emits sources with `title` and `url` on the `WebSearchDone` event.
224
226
 
227
+ Each `server_tool_use` block and its result block (`web_search_tool_result`, or `code_execution_tool_result` when a newer web search version runs the search from code execution) is kept on the assistant message as a `ServerToolPart` (see [Messages — Parts](../MESSAGES.md#parts)), and a streamed one also arrives as a `ServerToolDone` event. With `generate`, text blocks with citations keep their block too; a streamed turn keeps the text without its citations. Later turns send them back in place, so Claude reuses the results instead of searching again.
228
+
225
229
  ## Message Format
226
230
 
227
231
  The provider converts Riffer messages to Anthropic format:
228
232
 
229
- | Riffer Message | Anthropic Format |
230
- | -------------- | --------------------------------------------------------------- |
231
- | `System` | Added to `system` array as `{type: "text", text: ...}` |
232
- | `User` | `{role: "user", content: "..."}` |
233
- | `Assistant` | `{role: "assistant", content: [...]}` with text/tool_use blocks |
234
- | `Tool` | `{role: "user", content: [{type: "tool_result", ...}]}` |
233
+ | Riffer Message | Anthropic Format |
234
+ | -------------- | ------------------------------------------------------------------------ |
235
+ | `System` | Added to `system` array as `{type: "text", text: ...}` |
236
+ | `User` | `{role: "user", content: "..."}` |
237
+ | `Assistant` | `{role: "assistant", content: [...]}`, one block per part, in part order |
238
+ | `Tool` | `{role: "user", content: [{type: "tool_result", ...}]}` |
235
239
 
236
240
  ## Direct Provider Usage
237
241
 
@@ -259,6 +259,9 @@ Riffer::StreamEvents::ToolCallDone.new(
259
259
  Riffer::StreamEvents::ReasoningDelta.new("thinking...")
260
260
  Riffer::StreamEvents::ReasoningDone.new(part)
261
261
 
262
+ # Server tool blocks (if supported); see the Ordered Parts section
263
+ Riffer::StreamEvents::ServerToolDone.new(server_tool_part)
264
+
262
265
  # Web search (if supported)
263
266
  Riffer::StreamEvents::WebSearchStatus.new("searching", query: "search query")
264
267
  Riffer::StreamEvents::WebSearchDone.new(
@@ -343,6 +346,27 @@ yielder << Riffer::StreamEvents::ReasoningDone.new(part)
343
346
 
344
347
  If your adapter surfaces reasoning text but cannot yet replay it, call `yield_reasoning_done(yielder, text)` instead. It wraps the text in a `:text` part with no `format`, which persists for display and is skipped on replay.
345
348
 
349
+ ## Ordered Parts
350
+
351
+ By default the base class builds the assistant message's [parts](../MESSAGES.md#parts) from `extract_reasoning`, `extract_content` and `extract_tool_calls`, in that order. If your provider interleaves blocks in one turn and needs them back in that order, override `extract_parts` instead and return every block as a part, in response order:
352
+
353
+ ```ruby
354
+ def extract_parts(response)
355
+ response.blocks.filter_map do |block|
356
+ case block.type
357
+ when "text" then Riffer::Messages::Assistant::TextPart.new(block.text)
358
+ when "thinking" then Riffer::Messages::Assistant::ReasoningPart.new(type: :text, text: block.text, signature: block.signature, format: "my-provider-v1")
359
+ when "tool_call" then Riffer::Messages::Assistant::ToolCall.new(call_id: block.id, name: block.name, arguments: block.arguments)
360
+ when "search" then Riffer::Messages::Assistant::ServerToolPart.new(name: "web_search", kind: :call, id: block.id, data: block.to_json, format: "my-provider-v1")
361
+ end
362
+ end
363
+ end
364
+ ```
365
+
366
+ Then replay by walking `message.parts` rather than reading `content`, `tool_calls` and `reasoning` separately, and skip parts whose `format` you don't recognize.
367
+
368
+ For streaming, emit `ServerToolDone` for each server tool block, and keep text deltas, `ReasoningDone` and `ToolCallDone` in the order the blocks arrive. The agent loop builds the parts in that order.
369
+
346
370
  ## Reasoning Level
347
371
 
348
372
  When an agent sets [`reasoning`](../AGENTS.md#reasoning), riffer resolves the level to `model_options` params before calling `build_request_params`, so your provider receives ordinary options and never sees the level itself. It uses the model's [catalog](../CATALOG.md#reasoning) entry under your registration key first, then your provider's default. Override the private `default_reasoning_options` to supply one:
@@ -176,7 +176,7 @@ Gemini 3 models attach an opaque `thoughtSignature` to the first function call o
176
176
 
177
177
  Riffer keeps each signature on the assistant message as the `data` of an `:encrypted` [reasoning part](../MESSAGES.md#reasoning) tagged `format: "gemini-generate-content-v1"` (`Riffer::Providers::Gemini::REASONING_FORMAT`). A signature from a function call carries that tool call's `call_id` as its `id`; the text signature has no `id`. Streams emit them as `ReasoningDone` events. On later turns riffer puts each signature back on the part it came from. You don't have to do anything as long as the assistant messages stay in the history.
178
178
 
179
- A function call that Gemini didn't sign, such as one another provider made before the conversation switched to Gemini, goes out with Gemini's placeholder signature (`skip_thought_signature_validator`), which Gemini accepts in place of a real one. That covers the first call of each step when that call has no signature for this provider. Google warns that a call without its real signature may degrade the model's answer, so if you persist sessions, keep the `reasoning` key when you store messages.
179
+ A function call that Gemini didn't sign, such as one another provider made before the conversation switched to Gemini, goes out with Gemini's placeholder signature (`skip_thought_signature_validator`), which Gemini accepts in place of a real one. That covers the first call of each step when that call has no signature for this provider. Google warns that a call without its real signature may degrade the model's answer, so if you persist sessions, keep the reasoning when you store messages.
180
180
 
181
181
  ### Web Search
182
182
 
@@ -266,7 +266,7 @@ Gemini 3 models attach an opaque `thoughtSignature` to the first function call o
266
266
 
267
267
  Riffer keeps each signature on the assistant message as the `data` of an `:encrypted` [reasoning part](../MESSAGES.md#reasoning) tagged `format: "google-cloud-generate-content-v1"` (`Riffer::Providers::GoogleCloud::REASONING_FORMAT`). A signature from a function call carries that tool call's `call_id` as its `id`; the text signature has no `id`. Streams emit them as `ReasoningDone` events. On later turns riffer puts each signature back on the part it came from. You don't have to do anything as long as the assistant messages stay in the history.
268
268
 
269
- A function call without an Agent Platform signature goes out with Google's placeholder signature (`skip_thought_signature_validator`), which Agent Platform accepts in place of a real one. That happens when the call came from another provider, or from the Gemini Developer API, whose signatures Agent Platform rejects. It covers the first call of each step when that call has no signature for this provider. Google warns that a call without its real signature may degrade the model's answer, so if you persist sessions, keep the `reasoning` key when you store messages.
269
+ A function call without an Agent Platform signature goes out with Google's placeholder signature (`skip_thought_signature_validator`), which Agent Platform accepts in place of a real one. That happens when the call came from another provider, or from the Gemini Developer API, whose signatures Agent Platform rejects. It covers the first call of each step when that call has no signature for this provider. Google warns that a call without its real signature may degrade the model's answer, so if you persist sessions, keep the reasoning when you store messages.
270
270
 
271
271
  ### Web Search
272
272
 
@@ -96,6 +96,8 @@ provider.calls
96
96
  expect(provider.calls.last[:messages].last[:content]).to eq("Hi")
97
97
  ```
98
98
 
99
+ Each message is recorded as its `to_h`, so an assistant message has `parts` rather than `content` (see [Messages — Parts](../MESSAGES.md#parts)).
100
+
99
101
  ## Clearing State
100
102
 
101
103
  Reset stubbed responses:
@@ -96,7 +96,7 @@ Each `reasoning` item in a Responses API output becomes [reasoning parts](../MES
96
96
 
97
97
  The `:encrypted` part is always present, even when the response carried no `encrypted_content`, so the item's `id` is kept. When streaming, the parts are yielded as `ReasoningDone` events once the item completes; summary text still arrives as `ReasoningDelta` events before that.
98
98
 
99
- On the next request, parts sharing an `id` go back as one `reasoning` item, ahead of the assistant's text and function calls, in their original order and unchanged. This lets reasoning models carry their reasoning across turns and tool-call loops. You don't have to do anything as long as the assistant messages stay in the history; if you persist sessions, keep the `reasoning` key when you store them.
99
+ On the next request, parts sharing an `id` go back as one `reasoning` item, ahead of the assistant's text and function calls, in their original order and unchanged. This lets reasoning models carry their reasoning across turns and tool-call loops. You don't have to do anything as long as the assistant messages stay in the history; if you persist sessions, keep the reasoning when you store them.
100
100
 
101
101
  With `store` left on (the API default), OpenAI can resolve a replayed item from its `id` alone. With `model_options store: false`, the item can only be replayed from its encrypted payload, so request it explicitly:
102
102
 
@@ -118,6 +118,8 @@ model_options web_search: true
118
118
  model_options web_search: {search_context_size: "medium"}
119
119
  ```
120
120
 
121
+ With web search on, riffer also adds `"web_search_call.action.sources"` to the request's `include`, keeping any entries you pass, so `WebSearchDone` carries the sources OpenAI searched.
122
+
121
123
  ## Example
122
124
 
123
125
  ```ruby
@@ -69,6 +69,9 @@ groups:
69
69
  - source: TRACING.md
70
70
  slug: tracing
71
71
  description: OpenTelemetry spans
72
+ - source: STABILITY.md
73
+ slug: stability
74
+ description: What's safe to depend on and how upgrades work
72
75
 
73
76
  - label: Providers
74
77
  tone: jane