Skip to content

osmove/backlog-examples

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

backlog-examples

Working examples of agents driven by Backlog Cloud.

Backlog is the execution engine for AI agents: track work items, dispatch tasks, log every run, bill the tokens. These examples show what that looks like in practice — not "hello world", but actually-useful agents you could rip apart and adapt.

Why these exist

If you're building anything agentic, you've probably re-invented half of this already: a queue of things to do, status transitions, a way to record what the agent actually did, a budget. Backlog gives you one API for all of it. These examples show real workflows so you can decide if that API fits.

Quickstart

# 1. Get the CLI to create your account and a workspace
npm install -g backlog
backlog login              # creates an account on backlog.so
backlog ws create "demo"

# 2. Clone and configure these examples
git clone https://github.com/osmove/backlog-examples
cd backlog-examples
cp .env.example .env       # fill in your email/password and workspace id
npm install

# 3. Pick an example and run it
npm run agent:doc-bot:seed # creates demo work items in your workspace
npm run agent:doc-bot      # the agent picks one and works on it

Examples

A 90-line Node script that picks a "doc-fix" work item from your backlog, asks Claude (through the Backlog AI proxy) to add JSDoc comments to a local file, writes the annotated version to out/, and logs the whole operation as a Backlog Run with token usage and a diff artifact.

Showcases: listWorkItems, aiMessages, createTask, createRun, plus quota / billing visibility through getUsage.

Anti-features

These examples deliberately don't:

  • Bring their own LLM key. All AI calls go through Backlog's proxy so token usage is metered and capped per workspace.
  • Touch external services beyond Backlog and Anthropic. No GitHub PR creation, no Slack — that's left as an exercise so the example stays under 100 lines.
  • Try to be clever about prompts. The prompts are intentionally basic so you focus on the orchestration pattern, not on prompt-eng.

Self-hosted Backlog?

export BACKLOG_API_URL=https://backlog.your-company.com/api/v1

Same examples, your backend.

License

Apache-2.0 — go ahead and ship something with these as a starting point.

About

Working examples of agents driven by Backlog Cloud — see https://backlog.so

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors