Scan a repository and generate a .cercano/context.md file from key project files.
npx skills add https://github.com/bryancostanich/Cercano --skill cercano-init
Please help me install this Agent Skill. Skill: cercano-init Source: https://github.com/bryancostanich/Cercano/tree/main/plugins/skills/cercano-init Command: npx skills add https://github.com/bryancostanich/Cercano --skill cercano-init
Cercano projects often require a project-scoped knowledge context to guide local AI tooling. This skill automatically scans a repository to generate a concise context document that informs all Cercano tool calls, reducing repeated questions and boosting accuracy.
Run the cercano_init tool in your project root to generate a .cercano/context.md file.
High-intent search queries and answers about installing and using this skill.
To generate local AI project context, run the initialization tool in your project root. It scans key files like README, CLAUDE.md, .proto, .h, and config manifests to create a centralized .cercano/context.md file.
Project context prepending works by automatically attaching the generated .cercano/context.md file to the beginning of all subsequent Cercano tool calls, ensuring the local AI consistently receives your domain knowledge.
Yes, you need Ollama available. Initializing local AI project context requires a running Cercano server with Ollama accessible to scan your repository and extract domain knowledge.
The context initialization scans your project root for key files including README, CLAUDE.md, .claude/memory/*, .proto, .h files, and critical configuration manifests to extract domain knowledge.
Yes, you can merge host-provided context. The initialization tool reads your key files, extracts domain knowledge, and merges it with any host-provided context into the final .cercano/context.md file.
GoogleDevRelExplorationsScan a repository and write a .cercano/context.md file.
GoogleDevRelExplorationsPerform local web searches and synthesize citation-backed answers offline via Cercano.
trysmrReload user memory, project memory, and README content into a Markdown context block.
neuron-oneRun AI models locally with Ollama or llama.cpp for private, offline inference.
