./dtcw4 local installDacli
installDacli
About This Task
daCLI ("docs-as-code CLI") gives LLM agents structured access to your AsciiDoc and Markdown documentation — they can read the document structure, fetch a section, search, and edit content instead of grepping files by hand. It works in two modes:
-
a CLI (
dacli) for agents with shell access, and -
an MCP server (
dacli-mcp) for MCP-capable clients (Claude Code, IDEs, …).
This task installs daCLI and, if the project has an AGENTS.md or CLAUDE.md, adds a short conditional hint so agents know to use it (see Agent Hint).
Setup and Configuration
daCLI is a Python tool installed with uv.
If uv is not on your PATH, the task prints how to install it and stops — it does not fail with a stack trace.
|
Note
|
daCLI is installed straight from its Git repository (uv tool install git+https://github.com/docToolchain/dacli).
This is intentional — the PyPI package named dacli is an unrelated project.
|
To pin a specific ref (tag, branch or commit), set DTC_DACLI_VERSION:
DTC_DACLI_VERSION=v1.0.0 ./dtcw4 local installDacli
Running the MCP Server
After installation you can start the MCP server scoped to this project’s documentation (the task prints the exact command, using your configured inputPath):
dacli-mcp --docs-root src/docs
Register that command with your MCP client to let it navigate and edit the docs.
Agent Hint
When an AGENTS.md or CLAUDE.md exists, the task upserts a short, heading-anchored section into it (idempotent on re-run).
The hint is conditional: it points the agent at daCLI only when reading or editing the documentation, references your inputPath, and tells it to run dacli --help first.
If neither file exists, the hint is skipped and no agent file is created.
Further Reading and Resources
Source
View the source of this task: scripts/installDacli.groovy on GitHub
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.