Post-Mortem: Building a Local MCP Server for Codebase Memory using Ollama and ChromaDB

A developer built a local MCP server for codebase memory using Ollama and ChromaDB, enabling AI-assisted coding without sending proprietary code to cloud APIs. The project, zerikai_memory, supports local, cloud, and hybrid modes, with local mode running entirely on-device via Ollama. Benchmarks show mistral:7b runs faster on 8GB VRAM than ornith:9b, which offers better agentic coding performance but requires more memory.

Developers are pushing back against cloud API billing and the privacy risks of sending proprietary codebases to third-party endpoints. A Hacker News thread from this year put it plainly: the problem isn't the price per token, it's the unpredictability of usage-based billing when AI agents are continuously polling APIs. On Reddit, the privacy concern is starker -- for enterprise and defense work, sending company IP to OpenAI or Anthropic is a hard no regardless of cost. zerikai memory https://github.com/KikeVen/zerikai memory has a local mode for exactly this: everything runs through Ollama, nothing leaves your machine. We shipped mistral:7b as the default local model. ornith:9b dropped in June 2026, trained specifically for agentic coding tasks, so we tested both. Here is what we found. zerikai memory https://github.com/KikeVen/zerikai memory runs in three modes: cloud DeepSeek , local Ollama , and hybrid . Routing between them is handled by should use cloud at main.py:986 , a 4-step priority chain: explicit override, keyword match, word count threshold, then MEMORY MODE env var fallback. In local mode that function always returns false -- everything stays on-device. In local mode, every synthesis call hits query ollama at main.py:1555 . The model receives a project brief from load project context plus structured ChromaDB entity payloads: function signatures, file paths, line ranges, docstrings. It returns an answer with inline file:line citations. One call, no streaming, no tool loop. The retrieval layer ChromaDB + L2 distance + lexical rerank is model-agnostic. Both models received identical context in every query test. The only variable was synthesis. 8GB dedicated VRAM is the hard ceiling. Windows offers shared system memory as overflow, but inference on shared RAM through PCIe is slow enough to matter in practice. mistral:7b v0.3 -- 4.4GB, 32K context. Function calling via Ollama raw mode only: you construct AVAILABLE TOOLS prompts by hand and parse TOOL CALLS responses yourself. Last updated May 2024. ornith:9b-dense -- roughly 5.5GB estimated, built on Gemma 4 and Qwen 3.5. RL-trained for agentic coding tasks with native tool calling. 69.4 on SWE-Bench Verified, 43.1 on Terminal-Bench 2.1, matching or beating models 3x its parameter count. A standalone Python script hit Ollama directly with static ChromaDB payload samples constructed from real zerikai memory https://github.com/KikeVen/zerikai memory workspace entities. 3 queries, 3 samples each per model, raw latency measured at the HTTP layer. This test measures inference speed on the hardware, not synthesis quality against a live codebase. | Model | Mean | StdDev | Min | Max | |---|---|---|---|---| | mistral:7b | 6.14s | 3.58s | 2.92s | 14.57s | | ornith:9b | 13.39s | 5.76s | 8.77s | 25.67s | ornith's cold start on the first query hit 25.67s. It spilled into shared memory before Ollama pinned the model. Warmed samples settled at 9-17s. mistral fits clean in 8GB and runs at 3-7s warm. The benchmark script is in the repo in the blog/ directory . Run it against your own hardware before committing to a model swap. Both models were tested live through the zerikai memory universal-brain MCP layer against the real zerikai memory https://github.com/KikeVen/zerikai memory workspace index. Same 5 queries, models switched manually between runs, Ollama restarted between each switch. This test measures what a developer actually experiences in Cursor or VS Code -- not synthetic payloads, real ChromaDB retrieval. Same query, same ChromaDB context, same system prompt. Here is what came back. ornith:9b: Based on the retrieved codebase memory, here's how query routing works: Routing Logic: 1. Explicit Override: The should use cloud function first checks for an explicit use cloud parameter that can force cloud mode DeepSeek regardless of other conditions main.py:986-1005. 2. Keyword Detection: If the user query contains keywords like "architecture", "explain", "how does", or "overview", it routes to DeepSeek's cloud mode config.py:43-44. 3. Word Count Threshold: Queries exceeding 40 words automatically route to cloud/DeepSeek via should use cloud 's word count check main.py:992-994, with this threshold defined in config as CLOUD ESCALATION WORD COUNT = 40 config.py:37. 4. Default Mode: Falls back to whatever is set as DEFAULT MEMORY MODE , which defaults to "hybrid" unless explicitly changed via the MEMORY MODE environment variable config.py:31-32, with this function returning based on whether that mode equals "cloud" main.py:996-998. Important Note: The retrieved context does not contain explicit information about Ollama's role in the routing decision or how local model queries are constructed when cloud mode is NOT selected. mistral:7b: The query routing in zerikai memory decides between using Ollama or DeepSeek based on a 4-step priority chain. The priorities are as follows: 1. Explicit use cloud override. 2. CLOUD ESCALATION KEYWORDS keyword match. 3. CLOUD ESCALATION WORD COUNT length threshold. 4. DEFAULT MEMORY MODE fallback recommended . This decision-making process aims to balance privacy, speed, and API costs while providing maximum context nuance for the user. For more detailed information on modes, configurations, and usage, you can refer to the README.md file in the project repository. Sources: main.py:986, README.md:320, main.py:1495, main.py:1311, README.md:105 ornith:9b: The .brain directory path is defined and resolved through two key locations: Definition: In config.py , line 1-2, there's a constant that defines the base name for the brain directory config.py:1 : DB PATH = '.brain/' BRAIN DIR = Path DB PATH Resolution: In main.py , lines 120-145, within the init workspace function, this path is resolved by joining it with the workspace's absolute path main.py:130 : path = Path workspace path .resolve db path = path / DB PATH db path.mkdir parents=True, exist ok=True The resolution process: 1. Takes the provided workspace path parameter 2. Resolves it to an absolute path using .resolve 3. Joins it with the .brain/ directory constant 4. Creates the full directory structure if it doesn't exist mistral:7b: The DB PATH constant, which is used for defining the storage root for all workspace data, is defined in config.py : DB PATH = Path file .parent / ".brain" It resolves to zerikai memory/.brain/ . This path is platform-agnostic as the Path object in Python handles different operating systems' file paths. Sources: config.py:67, main.py:1034, main.py:818, README.md:375, main.py:1796 ornith:9b: Background brief synthesis avoids MCP timeouts by running as a fire-and-forget task launched with asyncio.create task , meaning background brief synthesis is called without awaiting or returning its result. This decouples the slow LLM-based brief generation from the MCP request lifecycle, so the IDE agent receives its response immediately rather than waiting for synthesis to complete. Sources: main.py:784 - 1.96 rerank main.py:1827 - 0.99 rerank main.py:393 - 0.98 rerank main.py:2094 - 0.94 rerank mistral:7b: Background brief synthesis avoids MCP timeouts by using asynchronous task creation with asyncio.create task in the background brief synthesis function. This approach allows the function to run independently without blocking the main event loop, ensuring timely responses and avoiding MCP timeouts. The synthesized brief is saved as a markdown file in .brain/contexts/<workspace id .md . Sources: main.py:784, main.py:1827, main.py:393, main.py:2094 ornith:9b: save to memory routes by file extension: files with .py , .js , .ts , .css , .html , or .md are processed via tree-sitter entity extraction the extract entities flow in code indexer.py . Other extensions fall through to DeepSeek/Ollama LLM summarization. For the tree-sitter path, tree-sitter parses the source and extracts one CodeEntity per function/method/class -- each holding signature, docstring/JSDoc, and metadata. These entities are upserted into ChromaDB with deterministic md5 IDs so re-scans overwrite duplicates , and token usage is logged to the SQLite tracker zerikai.db . Sources: main.py:1072, README.md:279, code indexer.py:169, code indexer.py:204, code indexer.py:30 mistral:7b: The save to memory function works by saving content to persistent vector memory in ChromaDB. It uses tree-sitter for code parsing for supported file types: .py, .js, .ts, .css, .html, and .md. Other formats fall back on DeepSeek/Ollama LLM summarization. It uses deterministic md5 IDs so re-scans overwrite duplicates and logs token usage to zerikai.db . Sources: main.py:1072, README.md:279, code indexer.py:169, code indexer.py:204, code indexer.py:30 Neither model had enough indexed context to answer this fully. What each model did with that gap is the most important data point in this test. ornith:9b: Based on the available codebase memory context, I cannot determine how asyncio semaphore specifically controls Ollama concurrency in local mode. The retrieved entities show: 1. should use cloud - Determines cloud usage main.py:986 2. MEMORY MODE configuration with cloud/hybrid modes config.py:30-45 3. background scan - Uses Semaphore 4 for tree-sitter and Semaphore 2 for LLM summarization during scanning main.py:1827 None of these entities contain the specific concurrency logic for query ollama at runtime. If this routing exists in the codebase, it's not present in the retrieved memory entries. mistral:7b: The project uses asyncio semaphore for controlling Ollama concurrency in local mode. The query can be answered by explaining that asyncio semaphore is used to manage the number of concurrent processes for tree-sitter parsing and LLM summarization during the scan loop, with a Semaphore 4 for tree-sitter parsing and Semaphore 2 for LLM summarization in local mode. Sources: config.py:62, config.py:53, main.py:1827, main.py:1555 mistral answered with confidence. The Semaphore 4 and Semaphore 2 values it cited are real -- but they belong to the background scan pipeline, not query-time Ollama concurrency. It applied the wrong context to a different question and presented the answer as fact. ornith said it didn't have the information and stopped. For an IDE assistant, a confident wrong answer costs more than an honest gap. The order of operations matters and it is not obvious from the README. After editing your .memignore file, run embedding-docstring on your codebase first. Then run scan workspace . If you scan first with sparse or missing docstrings, ChromaDB indexes thin vectors. Re-scanning won't fix it unless you re-enrich first and scan again. The memory is only as good as what tree-sitter extracted, and tree-sitter only extracts what is there. zerikai memory https://github.com/KikeVen/zerikai memory ships with the embedding-docstring skill for this reason. It audits and rewrites docstrings, comment blocks, and inline documentation across an entire workspace for vector embedding quality, covering Python, JavaScript, TypeScript, and HTML. It writes missing documentation from scratch and respects a .memignore file at the workspace root. The correct workflow is: .memignore → embedding-docstring → scan workspace → query Skip the first step and both models underperform. You will spend time blaming the model or the hardware when the real problem is what went into ChromaDB. Current status: works well with pi.dev, VS Code support in progress due to large file size constraints in some editors. Update: as of 7/14/2026 VS Code now supports large files, so the skill is usable in both Cursor and VS Code. As a secondary test, we compared briefs generated for the same workspace by DeepSeek cloud, sparse docstrings and ornith:9b local, after embedding-docstring enrichment . This is not a controlled comparison -- the docstring density differed between runs, so the model is not the only variable. What the comparison shows is that ornith:9b, given enriched ChromaDB context, produces dense, precise briefs: atomic overwrite semantics, naming convention breakdowns, explicit gap flags where documentation is missing. DeepSeek against sparse context produced thinner output with some inferred detail not present in the code. The takeaway is not that ornith beats DeepSeek for brief generation. It is that embedding-docstring enrichment is visible and measurable in the output. When the context is rich, ornith produces briefs good enough to feed meaningful synthesis queries. When it is not, neither model can compensate. Before this release, full local mode had a GPU saturation problem. synthesize deep brief at main.py:538 fired asyncio.gather across all 9 brief sections simultaneously with no concurrency gate. In local mode that meant 9 concurrent Ollama calls hitting the GPU at once -- guaranteed to saturate an 8GB card. The fix shipped alongside this test. A global ollama semaphore initialized in main.py after client setup gates build section calls through a build section safe wrapper when use cloud=False . Cloud and hybrid modes bypass the semaphore entirely -- DeepSeek handles its own rate limiting on the API side. ollama semaphore = asyncio.Semaphore OLLAMA MAX CONCURRENCY async def build section safe name : if not use cloud: async with ollama semaphore: return await build section name, workspace id, workspace path return await build section name, workspace id, workspace path OLLAMA MAX CONCURRENCY is configurable via .env , defaulting to 1 for 8GB hardware. Users on cards with more VRAM headroom can raise it. The ornith:9b brief in this post was generated with this fix in place -- full local mode brief synthesis is production-ready as of this release. If token pricing is the reason you are reading this, here is what a GPU upgrade costs against what you are spending on API calls: AMD cards RX 6700 XT 12GB, refurbished from $380 offer equivalent VRAM but require ROCm configuration. Ollama's CUDA path is plug-and-play on NVIDIA. AMD works but adds setup overhead. On 8GB RTX 3050 class , ornith:9b runs but cold starts are painful and VRAM headroom is tight. The RTX 3060 12GB is the practical sweet spot for local zerikai memory https://github.com/KikeVen/zerikai memory use. ornith:9b is the new default local model recommendation, replacing mistral:7b . On 8GB dedicated VRAM: ornith fits but runs tight. Cold start hits 25s when Ollama hasn't pinned the model. Warm synthesis at 9-17s is acceptable for a local-only workflow where you are not switching models or running concurrent GPU workloads. Set OLLAMA MAX CONCURRENCY=1 in .env . On 10-12GB dedicated VRAM RTX 3060 12GB or better : the model stays pinned, cold starts drop significantly, and citation precision is consistently better than mistral. Under 8GB dedicated VRAM, or if synthesis latency matters more than citation precision, use mistral:7b . Set OLLAMA MODEL=mistral:7b in .env . It handles synthesis correctly when context is dense. When context is thin, it will fill gaps with confident but wrong answers. The query test was clean and controlled. The model difference is real and attributable to ornith's training on agentic coding tasks, not hardware or docstring quality. Use the benchmark script in the repo to validate on your own machine before switching. 📖 Original Publication: This engineering post-mortem was originally published on the Zerikai Tech Blog. Read the clean, formatted web version at https://zerikai.com .