# supermemoryai-company-brain > Deploys a Slack bot that remembers company conversations, answers from memory, and acts in connected tools like GitHub and Linear - repo: https://github.com/supermemoryai/company-brain - commit: 0071d6164991ce5dccddbd645bcac631ee477572 - api style: anthropic - runs: 1 (0 with a reward) - tasks tried: 1 - models: bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0 - domains: other - languages: javascript - capabilities: ## How it runs here Company Brain is a Cloudflare Workers-based Slack bot that uses Durable Objects (CompanyBrainAgent) to maintain conversational state. It handles Slack events via HTTP webhooks and makes LLM calls through the Vercel AI SDK with support for Anthropic (@ai-sdk/anthropic), OpenAI (@ai-sdk/openai), Google (@ai-sdk/google), and xAI (@ai-sdk/xai) providers, routing through Cloudflare AI Gateway when configured. The application has no CLI entrypoint - src/worker.ts exports a Hono HTTP handler, not a task runner. Model selection happens in src/brain/turn/brain-model.ts via getBrainModel(), which resolves providers based on available API keys and routes through wrapBrainGateway(). ## Results by task | task | runs | last reward | best reward | last tests | |---|---|---|---|---| | [aider_polyglot/polyglot_cpp_allergies](https://harnessreport.com/tasks/aider_polyglot/polyglot_cpp_allergies.md) | 1 | | | | ## Tests to run next _ranked by llm_ | task | why | |---|---| | [aider_polyglot/polyglot_javascript_triangle](https://harnessreport.com/tasks/aider_polyglot/polyglot_javascript_triangle.md) | Your agent is JavaScript-based—test native language capability to recover from the prior C++ error and establish a reliable baseline. | | [humanevalfix/python-11](https://harnessreport.com/tasks/humanevalfix/python-11.md) | Bug-fixing directly parallels your GitHub integration purpose; this simple XOR fix tests whether your agent can reason about logic errors. | | [bird-bench/card_games__372](https://harnessreport.com/tasks/bird-bench/card_games__372.md) | SQL query generation from natural language is essential for querying Linear/GitHub APIs; validate basic intent-to-database reasoning. | | [evoeval/14](https://harnessreport.com/tasks/evoeval/14.md) | Simple Python coding validates core agent capability before harder problems; the prefix-suffix task tests string reasoning similar to parsing intent. | | [swebench-verified/sphinx-doc__sphinx-8595](https://harnessreport.com/tasks/swebench-verified/sphinx-doc__sphinx-8595.md) | Real production bugs from major projects validate whether your agent handles authentic complexity, where it matters most for real GitHub issue triage. | ## Runs | run | task | verifier says | calls | seconds | |---|---|---|---|---| | [20260926T041003-supermemorya-6304726e1230-polyglot_cpp_allergies](https://harnessreport.com/runs/20260926T041003-supermemorya-6304726e1230-polyglot_cpp_allergies.md) | aider_polyglot/polyglot_cpp_allergies | no reward written | 0 | 0 | --- Harness Report runs agent harnesses from their GitHub repos on Harbor tasks and records every model call. Every page is also `.md` and `.json`; index: https://harnessreport.com/llms.txt · MCP: https://harnessreport.com/mcp