Authentication
Waiting for extraction...
Expose docs over SSH so AI agents can verify before they answer
SSH-Docs runs a read-only SSH server over your docs directory. Agents can use ls, find, grep, and cat to inspect the latest documentation instead of guessing from stale memory.
Start in 3 Commands
pip install ssh-docs
ssh-docs serve ./docs --auth public
ssh localhost -p 2222 'cat AGENTS.md'Once started, SSH-Docs generates AGENTS.md, SETUP.md, and SKILL.md at your docs root so agents get explicit integration instructions.
What The Agent Gets
Agents receive plain command output they can cite directly in their reasoning.
$ pwd
/docs
$ ls
AGENTS.md
SETUP.md
SKILL.md
README.md
guides/
api/
$ grep -R "authentication" /docs
/docs/api/reference.md:## Authentication
/docs/guides/getting-started.md:Use bearer token authResult: exact file paths, exact lines, and no ambiguity about what is currently documented.
Live Agent Session
Animated example: an agent connects over SSH, navigates docs, and extracts concrete facts for its final answer.
Endpoints
Waiting for extraction...
Evidence Path
Waiting for extraction...
Coding Agent Exploration Replay
Alternative animation (Claude Code style): scan docs, open the right file, then extract answer-ready details with source path.
Summary: waiting for extracted findings...
Typical Agent Workflow
1. Discover
Use ls and find . -name "*.md" to map the docs quickly.
2. Validate
Use grep -R to find relevant topics and confirm exact section names.
3. Extract
Use cat, head, and tail to pull only the text needed for the task.
Available Shell Commands
pwd, ls, cd, cat, head, tail, find, grep, help, exit
Deployment Modes
Local Agent Development
Run ssh-docs serve ./docs --auth public for quick local usage and testing.
Team Internal Network
Bind publicly with --host 0.0.0.0 and keep access in your private network.
Production Endpoint
Use key auth and set --hostname docs.example.com so generated instructions are correct.