CodeGate v0.1.0

First release outside of the image release process.

Main changes are:

• Introduction of code based package selection (performance increase)
• SQLite-VEC replacement of weaviate (performance / stability increase)

What's Changed

• Add apiBase to autocomplete model by @eleftherias in #24
• Set pull policy to always in the compose file by @jhrozek in #38
• Non-Functional Requirements by @lukehinds in #40
• Implement Basic CI by @lukehinds in #48
• Run make format by @jhrozek in #65
• Implement provider interface and OpenAI and Anthropic providers by @jhrozek in #66
• mock uvicorn.run in CLI serve command tests by @lukehinds in #72
• Dependabot by @lukehinds in #71
• Bump actions/checkout from 3 to 4 by @dependabot in #73
• Bump softprops/action-gh-release from 1 to 2 by @dependabot in #74
• Bump actions/setup-python from 4 to 5 by @dependabot in #75
• Bump litellm from 1.52.14 to 1.52.15 by @dependabot in #78
• Bump actions/cache from 3 to 4 by @dependabot in #79
• Keep the code coverage high by @jhrozek in #80
• Module prompt by @lukehinds in #81
• Create github action for syncing and exporting vector DB by @yrobla in #69
• Add embedding model by @ptelang in #84
• Add inferencing models by @ptelang in #85
• requires-python was breaking make test by @jhrozek in #88
• Add inference code by @ptelang in #82
• Fix linting errors in CI by @lukehinds in #89
• Bump actions/checkout from 3 to 4 by @dependabot in #92
• Change the imports to absolute instead of relative by @aponcedeleonch in #86
• Specify the provider in the URL to properly route traffic by @aponcedeleonch in #96
• Include .gitattributes file for git lfs by @aponcedeleonch in #98
• Bump bandit from 1.7.10 to 1.8.0 by @dependabot in #99
• Bump litellm from 1.52.15 to 1.52.16 by @dependabot in #100
• feat: modify embedding method to use embedding class by @yrobla in #101
• Add input processing pipeline + codegate-version pipeline step by @jhrozek in #91
• Update config to add singleton instance and add inference unit tests by @ptelang in #105
• Update llama.cpp related code to use openai format by @ptelang in #107
• Add normalizer instead of abusing LiteLLM adapter by @jhrozek in #106
• Add a forgotten file by @jhrozek in #112
• feat: create Dockerfile for shipping codegate as image by @yrobla in #113
• Add a FIM pipeline to Providers by @aponcedeleonch in #102
• Make the readme more fun by @lukehinds in #125
• Vllm provider by @lukehinds in #124
• On-the-fly secret scanning by @lukehinds in #119
• Add formatting step to CI by @poppysec in #120
• If codegate is detected in the message, change the system prompt by @jhrozek in #117
• Bump ruff from 0.8.0 to 0.8.1 by @dependabot in #126
• Bump python from 3.12-slim to 3.13-slim by @dependabot in #127
• Normalize and denormalize llamacpp streaming reply by @jhrozek in #121
• fix: Logging module always producing JSON by @aponcedeleonch in #129
• Code snippet extraction pipeline step by @jhrozek in #130
• Add FIM functionalty for VLLM provider by @aponcedeleonch in #132
• Bump pytest from 8.3.3 to 8.3.4 by @dependabot in #138
• Bump weaviate-client from 4.9.4 to 4.9.5 by @dependabot in #137
• Bump dawidd6/action-download-artifact from 6 to 7 by @dependabot in #139
• Ollama provider by @lukehinds in #135
• Respond with JSON if the request is non-stream by @aponcedeleonch in #149
• feat: enable weaviate usage in codegate by @yrobla in #128
• Split the im_start/im_end tags that VLLM is using into individual messages by @jhrozek in #131
• Remove non-instruct qwen model by @ptelang in #151
• Use pipelines in Ollama provider by @aponcedeleonch in #152
• fix: remove internal weaviate_data folder by @yrobla in #153
• Wire up secrets pipeline by @lukehinds in #154
• Delete .github/workflows/release.yml by @lukehinds in #155
• Update prompts to fix rag by @ptelang in #157
• Database schema and sqlc configs by @lukehinds in #136
• Bump litellm from 1.53.1 to 1.53.2 by @dependabot in #164
• fix: update weaviate EmbeddedOptions import path by @lukehinds in #159
• Convert async _close_models in LlamaCppInferenceEngine by @lukehinds in #158
• Fix llamacpp streaming completion by @ptelang in #163
• Add an output pipeline that deobfuscated secrets by @jhrozek in #162
• feat: replace models with lfs pointer files, and fix testing by @yrobla in #166
• fix: directly add minillm model into the repo by @yrobla in #169
• feat: enable image builds for ci by @yrobla in #168
• Record the output of the LLM in the DB by @aponcedeleonch in #165
• fix: use regular workers for image builds by @yrobla in #171
• Normalize key messages for all providers by @aponcedeleonch in #170
• Silencing SQLAlchemy logs by @aponcedeleonch in #172
• Avoid DB conflicts when schema change by @aponcedeleonch in #173
• Make weaviate conform with codegate logging by @lukehinds in #174
• Bump weaviate-client from 4.9.5 to 4.9.6 by @dependabot in #184
• Bump aiosqlite from 0.19.0 to 0.20.0 by @dependabot in #183
• Bump docker/build-push-action from 6.7.0 to 6.10.0 by @dependabot in #180
• Bump docker/metadata-action from 5.5.1 to 5.6.1 by @dependabot in #179
• Bump docker/setup-buildx-action from 3.6.1 to 3.7.1 by @dependabot in #178
• Bump fastapi from 0.115.5 to 0.115.6 by @dependabot in #182
• Bump litellm from 1.53.2 to 1.53.3 by @dependabot in #181
• Added initial dashboard functionality by @aponcedeleonch in #175
• Add a modifying output pipeline step, use it to decorate code snippets by @jhrozek in #177
• update dependencies to fix docker images by @yrobla in #189
• Disable codegate system prompt by @jhrozek in #191
• Disable tools until we handle them properly by @jhrozek in #190
• Implement the /models endpoint for the VLLM provider by @jhrozek in #188
• Tune context and prompts by @ptelang in #200
• Record alerts as part of the pipeline by @aponcedeleonch in #197
• Obfuscate text in all messages, not just the last user one + display feedback on the secrets we obfuscated by @jhrozek in #193
• Add backup import packages by @yrobla in #203
• Add cryptocurrency wallet related regular expressions by @poppysec in #199
• feat: enable restore from previous backups by @yrobla in #204
• fix: remove backup folder before taking it again by @yrobla in #206
• Remove regular expressions for secret variable names by @poppysec in #210
• The Wallet Mnemonic regex is catching too many strings by @jhrozek in #208
• feat: expose an entrypoint to restore from backup by @yrobla in #211
• Further removal of overbroad regex by @poppysec in #212
• Add CORS Middleware to FastAPI by @aponcedeleonch in #213
• Extract packages in both input and output using the LLM the user called with by @jhrozek in #214
• Warn about malicious, deprecated or archived packages in output snippets by @jhrozek in #215
• Add FE code to Docker image by @aponcedeleonch in #207
• Warn less loudly about leaked secrets that were already redacted by @jhrozek in #220
• Add word "ecosystem" to the context by @ptelang in #226
• Modify storageclient to singleton pattern by @ptelang in #223
• Use def in FastAPI dashboard calls. by @aponcedeleonch in #222
• Fix backup by @yrobla in #230
• feat: add alerts sse notification endpoint by @peppescg in #227
• feat: allow to expose models base path and model name from cli by @yrobla in #231
• fix: download git models before publishing image by @yrobla in #232
• Reformat alerts on found packages by @aponcedeleonch in #235
• fix: deal with weaviate multiple network issues by @yrobla in #237
• Bump node from 20.18-slim to 23.3-slim by @dependabot in #229
• feat: allow to expose vllm_url as parameter in Docker by @yrobla in #240
• Generate snippet message only in case of bad packages are found by @ptelang in #236
• Filter Weaviate-matched objects by @ptelang in #233
• Add reponse format parameter to LLM chat completion call by @ptelang in #234
• fix: override translate method to return ChatCompletionRequest by @yrobla in #242
• fix: allow to disable the take/restore backup from import by @yrobla in #238
• feat: allow to parameterize urls in docker image by @yrobla in #245
• Implement no-stream handling for llamacpp by @ptelang in #251
• Add context message when no bad packages are found by @ptelang in #250
• FIM pipeline refactoring for Copilor by @jhrozek in #248
• fix: do not enclose urls with quotes by @yrobla in #255
• Modify package filtering to use to case insensitve matching by @ptelang in #256
• Add quotes around packages in the list by @ptelang in #258
• Use ollama python client for completion by @aponcedeleonch in #241
• Modify package extract prompt to clarify package is named entity by @ptelang in #259
• fix: lowercase the packages found in snippets by @yrobla in #260
• Update system prompt to perform security analysis by @ptelang in #264
• feat: add mountpoints for weaviate_data and models_path by @yrobla in #271
• Bump llama-cpp-python from 0.3.4 to 0.3.5 by @dependabot in #267
• Bump node from 23.3-slim to 23.4-slim by @dependabot in #266
• do not mount weaviate_data by @yrobla in #273
• Add regex for detecting AWS secret access key by @ptelang in #276
• feat: infer ecosystem from the user's query by @yrobla in #272
• Make sure we can speak with ollama localhosted from container by @aponcedeleonch in #275
• fix: correct urls for reporting packages by @yrobla in #277
• Move secret encryption step to the top of input processing by @ptelang in #278
• fix: copy default models to exposed volume on entrypoint by @yrobla in #283
• Link dashboard in message returned to the chat by @aponcedeleonch in #286
• fix: use node LTS version in Dockerfile by @peppescg in #287
• feat: import packages from external artifact by @yrobla in #288
• Create a codegate_volume directory. by @aponcedeleonch in #284
• feat: replace all data with just dummy data by @yrobla in #289
• CoPilot Implementation by @lukehinds in #268
• Bad merge , my bad by @lukehinds in #293
• fix: fix formatting issues by @yrobla in #294
• fix: fix all tests in main branch by @yrobla in #295
• fix: only show output about secrets redacted for the last message by @yrobla in #292
• Adds cli command to generate certificate by @aponcedeleonch in #291
• Bump node from 22-slim to 23-slim by @dependabot in #299
• Fix Unit Tests by @lukehinds in #298
• start_serving should be False by @lukehinds in #304
• We were using the non existent self.headers by @lukehinds in #308
• Send Generic Proxy Name by @lukehinds in #306
• Help proxy startup standout in logs by @lukehinds in #305
• fix(FE): update nginx conf for serving all routes by @peppescg in #309
• Improve SSL security of target connections by @lukehinds in #303
• Use tempdir rather then leave certs directory in place after running tests by @stackloke2e in #312
• Add FIM processing for Copilot by @jhrozek in #311
• Add copilot headers/auth for extracting package/ecosystem by @ptelang in #314
• Copilot chats are sent through an input pipeline by @jhrozek in #315
• Remove internal inference url by @lukehinds in #313
• Push to stacklok/codegate and sign / scan by @lukehinds in #317
• Deconstruct the streaming reply into chunks and send them back individually by @jhrozek in #316
• Inspect all user messages for malicious packages by @ptelang in #318
• Pipe the Copilot output chunks through the output pipeline by @jhrozek in #319
• Fix Image Build by @lukehinds in #320
• Notify about the number of secrets redacted since the last assistant message instead by @jhrozek in #321
• Fix port in message to match the documented port. by @aponcedeleonch in #324
• Bump ruff from 0.8.2 to 0.8.3 by @dependabot in #325
• Bump litellm from 1.54.1 to 1.55.0 by @dependabot in #326
• Bump weaviate-client from 4.9.6 to 4.10.1 by @dependabot in #327
• Bump docker/build-push-action from 5 to 6 by @dependabot in #328
• fix: readd tests that were added at cli command by @yrobla in #307
• Add logic to check if certs exist for generate_certs by @lukehinds in #334
• Modify the prompt to clarify that user's message can contain multiple… by @ptelang in #336
• Style updates for developer docs and README by @danbarr in #338
• Placeholder needed for UI Cert Download by @lukehinds in #341
• Copilot DB integration. Keep DB objects in context to record at the end. by @aponcedeleonch in #331
• Add Project MD files by @lukehinds in #347
• Pass in copilot extra headers when analyzing output packages by @jhrozek in #346
• Allows certificate download via the dashboard by @lukehinds in #352
• Create models directory under codegate_volume by @ptelang in #342
• A hotfix for the FIM pipeline by @jhrozek in #353
• simplify signature initialization for optimization by @lukehinds in #354
• Fix FIM pipeline with copilot by @jhrozek in #362
• Improve logging by @lukehinds in #370
• Reduce from info to debug by @lukehinds in #368
• Cancel the processing_task on connection_lost and on exception by @jhrozek in #375
• Cache FIM entries in memory to avoid repeated writes to DB by @aponcedeleonch in #372
• A temporary workaround to make Anthropic FIM working with Continue by @jhrozek in #377
• Avoid caching FE code at docker build by @aponcedeleonch in #378
• Add null check for output string before JSON parsing by @ptelang in #382
• Obfuscate secrets before sending a snippet out for analysis by @jhrozek in #332
• Update readme, contrib, and security guides by @danbarr in #386
• fix: do not instantiate dbreader on each call by @yrobla in #391
• fix: do not try to send buffer data if it is empty by @yrobla in #393
• Refine caching non-copilot FIM calls by @aponcedeleonch in #392
• Fixup minor string inconsistencies by @danbarr in #397
• Fix codegate-version support and pipeline error by @ptelang in #396
• docs: update README.md by @eltociear in #403
• Replace unpinned actions with pinned action by @stacklok-cloud-staging in #404
• feat: add custom logger that includes the origin by @yrobla in #406
• Fix a typo in mappings by @jhrozek in #407
• Bump uvicorn from 0.32.1 to 0.34.0 by @dependabot in #356
• Bump pytest-asyncio from 0.24.0 to 0.25.0 by @dependabot in #357
• Bump weaviate-client from 4.10.1 to 4.10.2 by @dependabot in #359
• Bump docker/setup-buildx-action from 3.7.1 to 3.8.0 by @dependabot in #412
• Bump actions/checkout from 4.1.7 to 4.2.2 by @dependabot in #413
• Bump pydantic-settings from 2.6.1 to 2.7.0 by @dependabot in #360
• Bump litellm from 1.55.0 to 1.55.4 by @dependabot in #414
• Remove unused cert_gen.py by @jhrozek in #410
• Fix confusing debug messages by @jhrozek in #409
• fix: FIM not caching correctly non-python files by @aponcedeleonch in #408
• Add initial set of integration tests by @ptelang in #411
• Add secret type and context lines to secret alert by @aponcedeleonch in #422
• fix: wrap provider error codes into HTTP Exceptions by @yrobla in #421
• Remove the need for secrets at Docker build by @aponcedeleonch in #426
• Add custom handler of exceptions by @aponcedeleonch in #425
• Tune the system prompts to extract package names or ecosystems by @jhrozek in #428
• improve TLS handling with SNI support and cert caching by @lukehinds in #432
• feat(doc): add codegate-ui documentation by @peppescg in #427
• Run filter using the specified packages/ecosystem by @ptelang in #434
• Revert "Remove the need for secrets at Docker build" by @lukehinds in #435
• Animated gif by @lukehinds in #441
• Bump click from 8.1.7 to 8.1.8 by @dependabot in #443
• Bump jinja2 from 3.1.4 to 3.1.5 by @dependabot in #448
• sqlite-vec vectorization database by @lukehinds in #438
• Clean up test duplication by @lukehinds in #447
• Bump litellm from 1.55.8 to 1.55.11 by @dependabot in #449
• Bump numpy from 2.2.0 to 2.2.1 by @dependabot in #452
• Remove package/ecosystem extraction using llm by @ptelang in #444
• Remove mistaken added DB file by @lukehinds in #450
• Update README.md by @lukehinds in #460
• Add system prompt only in case bad packages or secrets are found by @ptelang in #456
• Add rust package extraction and python aliased-import extraction by @ptelang in #458
• Automate sqlite-vec vector DB generation for images by @lukehinds in #455
• Vec port by @lukehinds in #465
• Fix import packages by @lukehinds in #466
• Change to image publish on release by @lukehinds in #467
• Fork with the correct arch in place by @lukehinds in #468
• Align flags with main app and rename files archive by @lukehinds in #469

New Contributors

• @eleftherias made their first contribution in #24
• @jhrozek made their first contribution in #38
• @lukehinds made their first contribution in #40
• @dependabot made their first contribution in #73
• @yrobla made their first contribution in #69
• @ptelang made their first contribution in #84
• @aponcedeleonch made their first contribution in #86
• @poppysec made their first contribution in #120
• @peppescg made their first contribution in #227
• @stackloke2e made their first contribution in #312
• @danbarr made their first contribution in #338
• @eltociear made their first contribution in #403
• @stacklok-cloud-staging made their first contribution in #404

Full Changelog: https://github.com/stacklok/codegate/commits/v0.1.0