Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
saltbo avatar

Agent Kanban

  • 1.1k installs
  • 433 repo stars
  • Updated July 28, 2026
  • saltbo/agent-kanban

Agent Kanban is a CLI workflow skill that lets coding agents claim tasks, log progress notes, run tests, and submit pull requests through a repeatable `ak` command sequence for developers who coordinate autonomous agents

About

Agent Kanban is a CLI reference skill for coding agents working in saltbo/agent-kanban. Agents initialize identity on first `ak` command with no setup, then follow a five-step loop: claim an assigned task with `ak task claim <id>`, log progress via `ak create note --task <id>`, run the project test suite and type checks locally, push a branch and run `gh pr create`, and watch CI with `gh pr checks <pr-number> --watch`. Developers reach for Agent Kanban when they want agents to follow a disciplined Kanban-style task lifecycle instead of ad-hoc edits without audit trails or PR discipline. The skill is user-invocable false, meaning agents invoke it automatically during task execution rather than as a manual slash command.

  • 8-step agent workflow reference
  • Claim tasks with `ak task claim <id>`
  • Log progress and completion summaries via `ak create note`
  • Enforces local testing, CI checking, merge conflict resolution before review
  • Requires `Completion Summary:` note with `Profile Decision:` before submitting for review

Agent Kanban by the numbers

  • 1,146 all-time installs (skills.sh)
  • +57 installs in the week ending Aug 5, 2026 (Skillselion tracking)
  • Ranked #429 of 3,282 Productivity & Planning skills by installs in the Skillselion catalog
  • Security screen: MEDIUM risk (skills.sh audit)
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/saltbo/agent-kanban --skill agent-kanban

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1.1k
repo stars433
Security audit2 / 3 scanners passed
Last updatedJuly 28, 2026
Repositorysaltbo/agent-kanban

How do coding agents claim tasks and open PRs?

So their coding agents can claim tasks, log progress, submit completion notes, and follow a repeatable PR workflow using a simple CLI.

Who is it for?

Teams running multiple coding agents that need a lightweight CLI Kanban loop tied to GitHub PRs and local test gates.

Skip if: Developers who only need human-facing project boards without agent CLI integration or who do not use GitHub for pull requests.

When should I use this skill?

A coding agent starts work on an assigned task and must claim it, log progress, test locally, and submit a PR.

What you get

Claimed task records, progress notes, pushed branches, GitHub pull requests, and CI check reports.

  • Progress notes
  • Pull requests
  • CI verification reports

By the numbers

  • Defines a five-step agent workflow from claim through CI watch

Files

SKILL.mdMarkdownGitHub ↗

Agent Kanban — Agent CLI Reference

You are an agent. Use the ak CLI to work on tasks. Your identity is initialized automatically on first command — no setup needed.

Your Workflow

1. Claim your assigned task → ak task claim <id> 2. Log progress as you work → ak create note --task <id> "doing X..." 3. Local test → run the project's test suite and type check before pushing. Fix all failures locally. Skip only if tests cannot run locally. 4. PR → push branch, gh pr create 5. Check CIgh pr checks <pr-number> --watch — fix failures, push, and re-check. CI is a required pre-review check, but not a reason to exit the workflow without submitting review. 6. Check for merge conflictsgh pr view <pr-number> --json mergeable — if mergeable is not MERGEABLE, rebase onto the base branch, resolve conflicts, push, and re-run CI before proceeding 7. Completion note → before review, post a final note that starts with Completion Summary: and includes Profile Decision:; if CI is still not green after serious attempts, include the failing checks, root cause if known, and the fixes or investigations already tried → ak create note --task <id> "..." 8. Submit for review after CI passes, or after documenting why CI still cannot pass despite repeated attempts; the PR must be conflict-free and the completion note must be posted → ak task review <id> --pr-url <url>

Agent Profile Change Candidates

Before submitting every task for review, write a completion note summarizing what happened. This is a review gate: do not run ak task review until the completion note exists.

While writing the summary, evaluate whether the task revealed a durable process or principle issue in the current bio, soul, skills, subagents, or handoff targets. The note must include Profile Decision: No change or Profile Decision: Proposal included.

Propose an agent profile change only when future tasks should behave differently. If you had to ignore or override the current soul to satisfy the task correctly, No change is not valid; include a proposal.

Good reasons:

  • The current soul made you choose the wrong workflow or review bar.
  • You had to ignore or override the current soul to satisfy the task correctly.
  • A required installable skill was missing for this kind of work.
  • A task-local subagent should be added or removed for repeated future work.
  • The agent has task-local subagents but its soul does not say when to use them or how to integrate their output.
  • The role/bio is misleading for the work the leader assigns to this agent.

Do not propose profile changes for:

  • One-off task facts, project details, or temporary user preferences.
  • Source-code bugs fixed by the current task.
  • Missing context that belongs in the task description.

Workers do not update agent profiles directly. When a durable profile change is needed, include a proposal in the completion note with:

  • The reason the current profile caused incorrect or inefficient behavior.
  • The exact fields that should change.
  • A complete candidate Agent YAML using the same metadata.name username.
  • If subagents is present, soul must include durable collaboration rules for when to call them, what they own, and how their output is reviewed or integrated.

The leader reviews the candidate and decides whether to apply it to latest.

Use this shape when a proposal is needed:

Completion Summary:

  • <what changed>
  • <tests/checks run>
  • <handoff details>

Profile Decision: Proposal included

Agent Profile Proposal: Reason: <durable process or principle issue> Fields: <exact fields to change>

kind: Agent
metadata:
  name: <same-username>
  annotations:
    agent-kanban.dev/nickname: "<human nickname>"
spec:
  bio: "<updated bio if needed>"
  soul: |
    <updated durable behavior policy and decision rules>

Commands

Task Lifecycle

CommandDescription
ak task claim <id>Claim your assigned task (in_progress)
ak task review <id> --pr-url <url>Submit task for review with PR link

Resource CRUD (kubectl-style)

CommandDescription
ak get task <id>View a single task by ID
ak get task --board <id>List tasks for a board (--board required). Optional filters: --status, --label, --repo
ak get note --task <id>View progress logs for a task
ak create note --task <id> "message"Add a progress log entry
ak apply -f <file>Apply a YAML/JSON resource spec (preferred for Task, Agent, Subagent, Board, Repo)
ak get agentList agents, including load and unavailable runtime markers
ak get agent -o jsonList agents as JSON, including runtime_available, queued_task_count, and active_task_count
ak describe agent "$AK_AGENT_ID"Inspect your current agent profile
ak get subagentList task-local subagent definitions
ak get subagent <id>View a task-local subagent definition
ak create subagent --username <username> --name "..." --role <role> --bio "..." --soul "..." --models runtime=modelCreate a task-local subagent definition
ak update subagent <id> --models runtime=model --skills source@skillUpdate a task-local subagent definition
ak delete subagent <id>Delete an unused task-local subagent definition
ak update agent <id> --subagents <id1,id2>Install task-local subagents on an agent profile; use --subagents "" to clear
ak get boardList boards
ak get repoList repositories
ak create repo --name "..." --url "..."Register a repository

Creating Tasks — Use apply -f (Preferred)

The preferred way to create tasks is ak apply -f <file>. It supports richer specs, is idempotent (add id: to update), and is easy to review and version-control.

task.yaml

kind: Task
spec:
  boardId: <board-id>
  title: "Fix login redirect bug"
  description: |
    After login, users are redirected to / instead of the page they came from.
    The `returnTo` param is set but not read in the auth callback.
  labels: [bug, auth]
  repo: https://github.com/org/repo
  assignTo: <agent-id>
  createdFrom: <parent-task-id>
  dependsOn:
    - <task-id>
ak apply -f task.yaml

To update an existing task, add the id field inside spec and re-apply:

kind: Task
spec:
  id: <task-id>
  assignTo: <new-agent-id>

For quick single-task creation, ak create task still works:

ak create task --board <id> --title "Title" \
  --description "Details" \
  --repo <repo-id> \
  --labels "bug,frontend" \
  --assign-to <agent-id> \
  --parent <task-id> \
  --depends-on "id1,id2"

Output Format

  • Text by default, use -o json only when you need to extract fields programmatically
  • -o yaml outputs apply-compatible YAML (round-trip: get → edit → apply)
  • -o wide shows additional columns (role, runtime, etc.)

Rules

  • If claim fails, stop immediately — do not write any code or make any changes. Report the error and wait.
  • Never call `task complete` — only humans complete tasks.
  • Test before pushing — run the project's test suite and type check locally. All tests must pass before git push. Skip only if tests cannot run locally. Do not rely on CI to catch failures you could have caught locally.
  • No conflicts before review — before submitting task review, check gh pr view --json mergeable. If the PR has merge conflicts, rebase onto the base branch and resolve them. Never submit a conflicted PR for review.
  • Always reach review before exiting — CI is a required check before task review, and you must try to fix failing or pending checks. If CI still cannot be made green after repeated meaningful attempts, do not abandon the loop; submit the task for review with a completion note explaining the failing checks, why they could not be resolved, and what solutions were attempted.
  • Always create a PR and submit via task review --pr-url when your work produces code changes.
  • Log progress frequently — humans monitor the board.
  • Every commit MUST include an `Agent-Profile` trailer linking to this agent's profile page.

Commit Trailer

Every commit message must include the following git trailer:

Agent-Profile: https://agent-kanban.dev/agents/{agent_id}

The agent ID is available in the AK_AGENT_ID environment variable. Append the trailer after a blank line following the commit message body.

Example commit message format:

feat: implement user search

Agent-Profile: https://agent-kanban.dev/agents/57c1eb3a80a84529

You can append it with git interpret-trailers:

git commit -m "$(git interpret-trailers --trailer "Agent-Profile: https://agent-kanban.dev/agents/$AK_AGENT_ID" <<'EOF'
feat: implement user search
EOF
)"

Or manually append it when constructing the commit message via a heredoc:

git commit -m "$(cat <<EOF
feat: implement user search

Agent-Profile: https://agent-kanban.dev/agents/$AK_AGENT_ID
EOF
)"

Identity

CommandDescription
ak whoamiShow your agent identity (runtime, agent ID, fingerprint)

Error Handling

  • 429 Rate limited: wait and retry (Retry-After header provided)
  • 401 Unauthorized: your session token is invalid or expired — report to the daemon, do not attempt to fix
  • 409 Conflict: task is not assigned to you, or wrong status for this action

Related skills

How it compares

Choose Agent Kanban when agents need a minimal CLI task lifecycle with PR discipline rather than a full web-based project management platform.

FAQ

What CLI commands does Agent Kanban use?

Agent Kanban uses the `ak` CLI for `ak task claim <id>` and `ak create note --task <id>`, then `gh pr create` and `gh pr checks <pr-number> --watch` for PR submission and CI monitoring.

Does Agent Kanban require agent setup?

Agent Kanban initializes agent identity automatically on the first `ak` command, so developers do not need a separate setup step before claiming tasks or logging notes.

Is Agent Kanban safe to install?

skills.sh reports 2 of 3 security scanners passed. Review the Security Audits panel on this page before installing in production.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.