openclaw-quire-skill

An OpenClaw skill that lets chat agents read tasks, projects, and task trees from Quire — and update existing tasks — by shelling out to the quire CLI.

v0.3 adds six gap-filler tools — four reads (list_subtasks, list_task_comments, list_statuses, list_tags) and two writes (set_task_dates, uncomplete_task) — bringing the curated surface to 13 reads and 6 writes. Heavier writes (delete, attach, move/transfer, comment edit, approval workflows) stay deferred; full CLI coverage is the future quire mcp server's job, not this skill's.

Install

openclaw skills install quire

View on ClawHub: clawhub.ai/quire/quire

Prerequisites

• quire CLI on $PATH.
• One-time auth: run quire login in a terminal — it opens a browser for OAuth. After that, every host using the skill reuses the same token store.

The skill declares quire in requires.bins, so OpenClaw will refuse to load it until the binary is present.

No environment variables, no API keys to paste.

What the skill exposes

Nineteen tools (thirteen reads, six writes), all backed by quire <command> --json:

Tool	Wraps	Kind
whoami	quire whoami	read
list_my_tasks	quire mine	read
list_project_tasks	quire task list	read
get_task	quire task get	read
list_subtasks	quire task subtasks	read
get_task_tree	quire task tree	read
search_tasks	quire task search	read
list_task_comments	quire comment list	read
list_projects	quire project list	read
get_project	quire project get	read
list_statuses	quire status list	read
list_tags	quire tag list	read
resolve_url	quire resolve	read
create_task	quire task create	write
update_task	quire task update	write
set_task_dates	quire task dates	write
complete_task	quire task complete	write
uncomplete_task	quire task uncomplete	write
add_comment	quire comment add	write

See SKILL.md for the full manifest the model reads at install time — including when-to-use guidance, common pitfalls, and example prompts.

Why a skill and not a hosted integration?

Because quire-cli already exists, already handles auth, and already speaks JSON. The skill is ~one Markdown file telling the model which command to run for which question. There is nothing to host, nothing to scale, nothing to keep secret.

The strategic plan is for quire-cli itself to expose an MCP server in a future version (quire mcp). When that lands, this skill stays useful for OpenClaw-specific contexts but new MCP-aware clients can talk to Quire directly.

Development

The skill is plain Markdown — no build step, no tests beyond manifest validation. CI runs the validator on every PR (see .github/workflows/validate.yml); run it locally with:

bash .github/scripts/validate.sh

The validator parses every quire ... invocation out of SKILL.md's bash code fences and asserts each one resolves against the installed CLI's --help output. If you add a new tool section, the CI catches it automatically — no separate manifest to keep in sync.

Publishing to ClawHub (requires clawhub login first):

clawhub skill publish . --slug quire --owner quire --version <semver>

Bump <semver> for each release — the registry rejects republishing an existing version. Add --changelog "..." to annotate the release; the narrative changelog lives in CHANGELOG.md.

License

MIT-0 — no attribution required.