Skip to content

Build AI methods with your coding agent, then run them anywhere: as an MCP for chatbots, as a webapp for people, or via API for your software.

Merge branch 'main' into dev

Louis Choquelad1319a5fb29
diff --cc CHANGELOG.md
index f8ac987c5,d26e38fbb..f2ffa25f7
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@@ -1,131 -1,10 +1,136 @@@
- # Changelog
+ ## [v0.17.6] - 2026-02-14
+
+ ### Added
+
+ - **Claude Code GitHub Actions**: Added `claude.yml` workflow for interactive Claude Code assistance on issues and PR comments, and `claude-code-review.yml` workflow for automated code review on pull requests.
+
+## Unreleased
+
+**Highlights:**
+
+- **`PipeCompose` Construct Mode** — New mode for deterministically building `StructuredContent` objects without an LLM. Compose fields from working memory variables, fixed values, templates, and nested structures.
+
+- **Content Storage System** — Configurable storage for generated artifacts (images, extracted pages). Supports `local` filesystem (`.pipelex/storage/`), `in_memory`, **AWS S3** (`pip install pipelex[s3]`), and **Google Cloud Storage** (`pip install pipelex[gcp-storage]`). Cloud providers support both public URLs and time-limited signed URLs. Content referenced via stable `pipelex-storage://` URIs.
+
+- **Hugging Face Backend** — Support for Hugging Face Inference API, including `qwen-image` text-to-image model.
+
+- **Python 3.14 Support** — Officially tested and supported.
+
+- **Pipelex Gateway** — The deprecated `pipelex_inference` backend is now replaced by `pipelex_gateway`, featuring remote model configuration fetching so you always have access to the latest models without updating Pipelex.
+
+ **Getting your API key:**
+
+ 1. Join our Discord community at [go.pipelex.com/discord](https://go.pipelex.com/discord)
+ 2. Request your free API key (includes credits for testing and development)
+ 3. Add it to your `.env` file: `PIPELEX_GATEWAY_API_KEY=your-key-here`
+
+ **Accepting the Terms of Service:**
+
+ When you run `pipelex init`, you'll be prompted to accept the Gateway terms of service. By using Pipelex Gateway, telemetry is automatically enabled and identified by your API key (hashed for security) to monitor service quality and enforce fair usage. **We only collect technical data** (model names, token counts, latency, error rates)—never your prompts, completions, or business data. See our [Privacy Policy](https://go.pipelex.com/privacy-policy).
+
+ **⚠️ Migration deadline:** If you were using `pipelex_inference`, please migrate soon—the legacy service will be shut down within a week. Get your new Gateway key at [app.pipelex.com](https://app.pipelex.com/).
+
+- **Pydantic Structure Generation** — Two new CLI commands bridge Pipelex's declarative concepts with your Python code:
+
+ **`pipelex build structures /library_dir/`** — Generates Pydantic models from all concept definitions found in the specified library directory. Now you have your structures as Python code: you can iterate on them, add custom validation functions, or use them as type hints in your code.
+
+ **`pipelex build runner`** — Now automatically generates both the Python runner file AND the required Pydantic structures. When you run this command, it creates a complete, ready-to-execute Python script that imports the generated structures, so you can immediately use typed objects in your pipeline code.
+
+ See the [Build Commands documentation](https://docs.pipelex.com/home/9-tools/cli/#build-commands) for usage examples.
+
+- **Langfuse & OpenTelemetry Observability** — New OpenTelemetry-based observability system enables powerful tracing and Evals through Langfuse integration. Also supports OTLP-compatible backends (Datadog, Honeycomb, etc.). Configured via `.pipelex/telemetry.toml`.
+
+### Added
+
+- New backends & models
+ - Google `gemini-3.0-flash-preview`
+ - Mistral OCR models `mistral-ocr-2503`, `mistral-ocr-2505`, `mistral-ocr-2512`
+ - **Scaleway** inference provider support for open-source models
+ - **Mistral** Document AI (OCR) via Pipelex Gateway (disclaimer: it's flaky, needs many retries via tenacity)
+ - **Portkey AI** backend integration for unified access to multiple models through a single API key
+ - Mistral Document AI (OCR) via Pipelex Gateway (disclaimer: it's flaky, needs many retries via tenacity)
+- **Unified URI Handling System**: New `pipelex.tools.uri` module providing type-safe parsing for HTTP/HTTPS URLs, local file paths, file URIs, `pipelex-storage://` URIs, and base64 data URLs.
+- **Automatic Data URL Normalization**: Pipeline pre-processing step that converts large `data:` URLs in `ImageContent` to `pipelex-storage://` URIs for improved performance. Configurable via `is_normalize_data_urls_to_storage` in `pipelex.toml`.
+- **`PreparedImage` Abstraction**: New models (`PreparedImageHttpUrl`, `PreparedImageBase64`) representing images ready for LLM provider APIs.
+- Comprehensive unit and integration tests for URI resolver, data URL normalization, and `pipelex-storage://` image handling.
+- **Cloud Storage Providers**: Added support for AWS S3 (`pip install pipelex[s3]`) and Google Cloud Storage (`pip install pipelex[gcp-storage]`), both supporting public URLs and time-limited signed URLs. These can now be used to store images generated by image generation models, images extracted from PDFs and any image you provide as input. Thanks to these, we now manipulate http URLs which are way more efficient for API calls and distributed computing, compared to base64 data URLs. Configured in `pipelex.toml` under the `[pipelex.storage_config]` section, using a top-level `method` setting with provider-specific tables (e.g., `[pipelex.storage_config.local]`, `[pipelex.storage_config.s3]`).
+- Agent rules: Added `pipelex_standards.md` outlining standards for the Pipelex configuration system, also included as rules for AI development agents.
+- Pipelex Gateway service with terms of service management via `.pipelex/pipelex_service.toml` and interactive acceptance flow in `pipelex init`
+- Configurable retry logic with exponential backoff for inference API calls, configurable in `pipelex.toml` under `[cogt.tenacity_config]`
+- Context manager support for the `Pipelex` class (`with Pipelex.make(): ...`) for graceful shutdown
+- Validation for Pipelex Bundle concept keys: cannot create a native concept
+- Validation for `PipeSequence`: The multiplicity of the output of the sequence must be the same as the multiplicity of the output of the last step.
+- Validation for `PipeFunc`: The multiplicity of the output of the pipe must be the same as the multiplicity of the output of the function: If the multiplicity is true, the return type of the function must be a subclass of `ListContent`. If the multiplicity is false, the return type of the function must not be a subclass of `ListContent`.
+- **Pipelex Gateway Model Management**: New CLI commands (`pipelex-dev update-gateway-models`, `pipelex-dev check-gateway-models`) and corresponding `make` targets (`ugm`, `cgm`) to generate and verify gateway model catalog. CI now validates this documentation is up-to-date.
+- `pipelex init agreement` command: Allows accepting Gateway terms of service without resetting entire configuration.
+- Test suite pre-flight check: Verifies Gateway terms acceptance before running tests, providing clear error messages.
+- "Gateway Available Models" documentation page: Auto-generated reference of all LLM, Document Extraction, and Image Generation models available through the Gateway.
+
+### Fixed
+- Fixed a bug with the validation of the multiplicity of the output of the `PipeCondition`.
+- Fixed a bug with the file names of the structures generated by the `pipelex build structures` command.
+
+### Changed
+
+- **Breaking**: User override config renamed from `pipelex_super.toml` to `pipelex_override.toml`.
+- **Image Generation Architecture**: Refactored to taxonomy-based approach. Standardizes parameter translation (`aspect_ratio`, `quality`, `output_format`) to provider-specific APIs.
+- **Content Handling Overhaul**: `GeneratedImage` replaced by internal `GeneratedImageRawDetails`; `ImageContent` is now standard (without `base_64` field). `PipeExtract` outputs `PageContent` list directly. Content persistence now handled automatically by storage system.
+- **Document Extraction Improvements**: `pypdfium2` extractor now extracts embedded images from PDFs. Response parsing uses dedicated Pydantic schemas for validation.
+- **Default Model Change**: `extract_text_from_visuals` deck now defaults to `azure-document-intelligence` instead of `mistral-ocr`.
+- **`pipelex_inference` replaced by `pipelex_gateway`** with remote model configuration fetching for always-current model support
+ - New environment variable `PIPELEX_GATEWAY_API_KEY`
+ - Default routing profiles updated from `pipelex_first` to `pipelex_gateway_first` with automatic migration in `pipelex init`
+- Telemetry system split into two streams:
+ 1. Pipelex Gateway telemetry for service monitoring (never collects prompts/completions/business data)
+ 2. Custom telemetry to user-configured backends
+- **Telemetry configuration restructured** in `telemetry.toml`:
+ - Renamed `[posthog]` to `[custom_posthog]` to distinguish user's PostHog from Pipelex Gateway telemetry
+ - Added new `[custom_portkey]` section with `force_debug_enabled` and `force_tracing_enabled` settings (replaces `FORCE_PORTKEY_DEBUG` and `FORCE_PORTKEY_TRACING` environment variables)
+- **Main configuration defaults updated** in `.pipelex/pipelex.toml`:
+ - Most settings now commented out by default (shown as examples rather than active overrides)
+ - `is_generate_cost_report_file_enabled` default changed from `true` to `false`
+ - `pipelex_override.toml` (final override) moved from repo root to `.pipelex/` directory
+ - `telemetry_override.toml` (personal telemetry settings) moved from repo root to `.pipelex/` directory
+- **Image Prompt Representation**: Redesigned `PromptImage` models—consolidated `PromptImagePath` and `PromptImageUrl` into `PromptImageUri`; now uses Pydantic discriminated union for URI, base64, and raw bytes sources.
+- **Centralized Image Preparation**: Moved image fetching and base64 conversion logic to `pipelex.cogt.image.prompt_image_utils`, simplifying LLM provider plugins (Anthropic, Google, Mistral, OpenAI).
+- **Unified Resource Loading**: Updated all file/URL reading components (PDF renderers, document extractors) to use the new URI handling system, replacing the legacy `pipelex.tools.misc.path_utils` module.
+- **Async HTTP Fetching**: Renamed `fetch_file_from_url_httpx_async` to `fetch_file_from_url_httpx`; removed redundant synchronous version.
+- Documentation: clarified **Setup (first run)** vs **Configuration (TOML reference)**, added a Setup overview page, and added contributor docs for configuration defaults/overrides.
+- `pipelex init` now creates a documented `telemetry.toml` template instead of prompting for preferences
+- Model catalog updated with latest models (gpt-5.1, claude-4.5-opus, gemini-3.0-pro, etc.) and updated waterfalls in `base_deck.toml`
+- Model constraints refactored from simple lists to structured `valued_constraints` dictionaries (e.g., `valued_constraints = { fixed_temperature = 1 }`)
+- New OpenAI Responses API implemented—now differentiates between `openai_completions` and `openai_responses`
+- CLI commands refactored to use centralized `Pipelex` initialization factory for improved error handling
+- `pipelex doctor` command enhanced to detect outdated `telemetry.toml` formats and suggest fixes
+- The `runner`, `structures`, and `inputs` CLI commands now accept `--output-dir` option (defaults to `target_dir`)
+- Cost report now displays a note clarifying that it only includes LLM costs
+- `description` field is not Optional anymore in the `PipeAbstract`, `PipeBlueprint` and `PipeSpec` classes.
+
+### Removed
+
+- **`openai_utils` Module**: Removed `pipelex.plugins.openai.openai_utils`; logic now in centralized image preparation utilities.
+- **Pipeline Tracking feature**: Removed entirely, including the `pipelex/pipeline/track` module, `PipelineTracker` components, related configuration, tracker calls in pipe controllers, and associated documentation.
+
+### Deprecated
+
+- `pipelex_inference` backend in favor of `pipelex_gateway` (marked as "🛑 Legacy" in configuration template)
+
+### Migration Notes
+
+- **Telemetry configuration migration**: If you have an existing `telemetry.toml`, rename:
+ - `[posthog]` → `[custom_posthog]`
+ - `[posthog.tracing]` → `[custom_posthog.tracing]`
+ - `[posthog.tracing.capture]` → `[custom_posthog.tracing.capture]`
+ - Or run `pipelex init telemetry --reset` to regenerate the file with the new structure
+
+### Refactored
+
+- `ConceptFactory.make_from_blueprint` method now correctly handles native concepts
+- The output (and inputs) of a pipe is now a `StuffSpec` object instead, that holds the concept and the multiplicity.
+- Renamed `domain` into `domain_code` when relevant.
+- Refactored the dry run methods of the `PipeAbstract` class.
+
## [v0.17.5] - 2026-01-16
- Added target `make docs-deploy-404` to deploy the 404.html file to the gh-pages root for versionless URL redirects.