Skip to content

Guide · Agent skill

How to use the Refresh Product Status agent skill (refresh)

Updated · written with AI from the project's own documentation and checked against our catalogue data

The refresh skill is an agent skill for the Agno company framework that keeps a project-tracking status page for a Product up to date using current work and the Product's configured source documents. This guide explains what the skill makes an agent do, how to install it, how to ask for it in plain language, and what to check before you rely on its output. By the end you will know how to set it up, trigger a refresh, and review the proposals it produces.

Refresh Product Statusagno-agi/agno · Apache-2.0

Quick start

  1. 1Install the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills.
  2. 2Confirm the Agno requirements. Make sure the Agno company/Product tools (refresh_company_product, read_product_source, intake_product_source) are available, and that the Product has a playbook and configured Sources.
  3. 3Ask for a refresh. Name the Product in a plain request, such as asking the agent to refresh its status.
  4. 4Review Project completion proposals. Read the proposals saved in the wiki by refresh_company_product and decide which to accept.
  5. 5Accept approved Projects through the API. Accept approved Project completions through the Project transition API. Do not let the agent transition them directly.
  6. 6Check status.md. Confirm that status.md reflects the observed progress and failures, and that archived work has been removed.

What the refresh skill makes the agent do

The skill is a set of written instructions (a SKILL.md file) that teaches an agent how to refresh the status of a Product in the Agno company wiki. The agent reads the Product named in your message and its playbook, then calls the refresh_company_product tool. That call applies the Product's archive window (metadata.archive_window_days, which defaults to 14) and returns Project completion proposals saved in the wiki.

The agent then reads the remaining Projects and Tasks, inspects the configured Sources with read_product_source, and records observed progress and failures in status.md. Removing references to archived work from status.md keeps old items out of future agent context.

  • It proposes Project completions but never completes a Project.
  • It proposes Task changes but never accepts work or marks a Task Ready or Done.
  • If nothing has changed, it returns a quiet successful check without writing duplicate pages.
  • If maintenance fails, it reports the failure instead of claiming success.

Install the refresh skill

You can install the skill in either of two ways. Download the ZIP from this page and unzip it into one of these locations: ~/.claude/skills/ for your personal skills, or .claude/skills/ inside a project folder for skills shared with that project.

Alternatively, upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

  • Personal install: unzip into ~/.claude/skills/
  • Project install: unzip into the project's .claude/skills/ folder
  • App install: Settings → Capabilities → Skills, then upload the ZIP

Check the requirements before you start

The skill only works inside the Agno framework. It depends on the company/Product tools refresh_company_product, read_product_source and intake_product_source, so it will not work on its own. It also needs a configured Product playbook and configured Sources in the Agno company wiki.

The skill does not run scripts, but it does use the network, because it reads configured sources.

  • Agno framework with the company/Product tools (refresh_company_product, read_product_source, intake_product_source)
  • A configured Product playbook and Sources in the Agno company wiki
  • Runs scripts: no
  • Uses network: yes

Trigger the skill with natural requests

You do not need a special command. Name the Product in your message and ask for a status refresh. The agent reads the Product identified in the message and follows the skill's steps from there.

Keep requests specific to one Product. The skill works on a single Product at a time, as described in its instructions.

  • "Refresh the status of the Website Redesign product."
  • "Update the status page for the Mobile App product from its current work and sources."
  • "Run the maintenance check on the Billing product and tell me what changed."

Review the proposals before accepting them

The skill is built to propose, not decide. When it returns Project completion proposals, present them to a human who accepts them through the Project transition API. The agent is instructed never to transition Projects directly.

Task proposals are sent with intake_product_source. The skill notes that propose_company_task is unavailable to Routines, so Task proposals go through that intake tool.

  • Read each Project completion proposal before accepting it.
  • Accept Projects only through the Project transition API.
  • Treat the status.md update as a summary of observed evidence, and check it against the source documents.

Understand how the skill reads sources

For each Source, the agent first calls read_product_source without a path to get a paged, content-free index. It can use pattern and next_offset to narrow discovery, then pass a returned path to read one document. Large documents are read in parts using next_content_offset.

The agent compares the full-document content_hash across fragments and rereads a document if the hash changes. It never infers content from an index entry alone.

  • Index first: read_product_source with no path
  • Narrow with pattern and next_offset
  • Read one document by passing a returned path
  • Continue large documents with next_content_offset
  • Reread if content_hash changes

Know the safety limits

The skill reads configured sources and writes their observed content into status pages. If those sources hold sensitive material, that content can end up in the wiki. Make sure sources are appropriately restricted before enabling the skill.

Maintenance only deletes recognized generated pages older than the archive window. The skill keeps brief.md and status.md.

  • Restrict configured Sources to content that is appropriate for status pages.
  • Expect deletions limited to recognized generated pages older than the window.
  • Do not use the skill to close work without review.

What people use it for

  • Keeping a wiki-style status page for a Product current without manual edits after every change.
  • Running a scheduled maintenance routine that refreshes Product status and reports failures.
  • Preparing an evidence-based status summary that a project manager reviews before accepting changes.
  • Clearing archived Projects and Tasks out of agent context so old items stop appearing.

Frequently asked questions

Is the refresh agent skill free?

The skill is published under the Apache-2.0 licence, as listed for the agno-agi/agno repository. Check the licence file in the repository for the full terms before you use or redistribute it.

What do I need to install the refresh skill?

You need the Agno framework with the company/Product tools refresh_company_product, read_product_source and intake_product_source. You also need a configured Product playbook and Sources in the Agno company wiki. The skill will not work on its own.

Where do I put the SKILL.md file?

Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

Does the refresh skill close Projects automatically?

No. The skill proposes Project completions and saves them in the wiki, but it never completes a Project. A human must accept each proposal through the Project transition API.

Can I use this skill outside Agno?

No. The skill depends on Agno-specific tools, so it is not intended for use outside the Agno company framework.

Why does the agent not show my source content?

The agent reads the full document before using its content and never infers content from an index entry. If the content_hash changes between fragments, it rereads the document. Check that the Source is configured and reachable in the wiki.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use the refresh skill if you run the Agno company/Product framework and want status pages refreshed from real work and source evidence, with a human approving every completion and Task change. Pick something else if you need a standalone tool, fully autonomous project management, or if your sources hold material you cannot expose in status pages.

Download the skill