Guide · Agent skill
How to use the Report Save and Link Check agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The reports skill gives an AI agent a consistent way to save recurring output, such as morning briefings, inbox sweeps, social mentions and meeting digests, as dated markdown files, and to find the latest one again by a simple keyword. It also adds a link check that runs before a report is delivered. This guide explains what the skill does, how to install it, how to ask for it in plain language, and what it needs before you rely on it.
Report Save and Link Checkgarrytan/gbrain · MIT Download the skill Quick start
- 1Download and unzip. Download the skill ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
- 2Or upload in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the skill.
- 3Confirm requirements. Check that the gbrain CLI is installed, your brain repository is under git, and you can access skills/_output-rules.md.
- 4Ask for a save. Ask the agent to save a report. It writes to reports/{category}/{YYYY-MM-DD-HHMM}.md with frontmatter.
- 5Load by keyword. Ask for the latest briefing or email report by keyword, and the agent finds the newest matching file in that category.
- 6Run the link check. Before sending a report with links, ask for a link quality check. The agent fixes or removes Broken, Dead and Indirect links.
What is an agent skill?
An agent skill is a folder containing a SKILL.md file that tells an AI agent how to handle a specific kind of task. When the skill is installed and a request matches its triggers, the agent follows the instructions in that file.
This skill is named reports. Its SKILL.md is the only file in the package reviewed for this page.
What the reports skill makes the agent do
The skill covers three jobs: saving reports, loading the latest report, and routing a keyword to the right report category.
Saving writes each report to a path shaped like reports/{category}/{YYYY-MM-DD-HHMM}.md with frontmatter that includes a title, type, report_type, category, date and time. Loading finds the most recent report file for a given category. Keyword routing maps common words to categories: "email" or "inbox" goes to ea-inbox-sweep, "social" or "mentions" goes to social-mentions, "briefing" or "morning" goes to morning-briefing, and "meeting" goes to meeting-sync. Custom mappings can be configured.
Before an outbound report reaches a user, the skill runs an Actionability Gate. The agent checks that each link lets the user click and act immediately. It flags four failure modes: Broken (fake or placeholder links), Dead (links to brain pages not yet committed and pushed), Indirect (a profile, homepage or channel standing in for specific content) and Missing (an actionable item with no link). Broken, Dead and Indirect links block delivery. Missing links produce a warning.
- Save: reports/{category}/{YYYY-MM-DD-HHMM}.md with frontmatter
- Load latest: returns the newest report in a category
- Keyword routing: "briefing" finds the morning briefing, "email" finds the inbox sweep
- Link gate: checks links before delivery, with up to 2 retries
How to install the reports skill
Download the ZIP from the page where you found this skill. Unzip it into one of two places: ~/.claude/skills/ to make the skill available for your user account, or .claude/skills/ inside a specific project to limit it to that project.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into .claude/skills/ in your project folder
- Claude apps: Settings → Capabilities → Skills, then upload
Requirements before you use it
The skill depends on the gbrain CLI and a brain repository tracked with git. Saving, searching and committing reports all go through these tools. The link check also relies on git status and gbrain get commands.
The link rules are referenced from skills/_output-rules.md, which is not included in the files reviewed. Without access to that file, the link canon the skill points to is unavailable.
- gbrain CLI installed
- A brain repository under git
- Access to skills/_output-rules.md for the link rules
How to trigger the skill with natural requests
You do not need exact command names. The skill's triggers include phrases such as "save report", "load latest report", "what's the latest briefing", "show me the pulse", "report quality", "link quality check" and "validate report links". Keyword routing also lets you ask for a category by a loose word such as "email" or "meeting".
Be specific about the category when you can. Asking for "the latest briefing" routes to morning-briefing, while "the inbox report" routes to ea-inbox-sweep.
- "What's the latest briefing?"
- "Load the latest email report."
- "Save this as a report."
- "Run a link quality check on this report before I send it."
- "Validate report links."
Safety notes: scripts and network
The skill is listed as not running scripts and not using the network. There is no enforcement script: the agent runs the link checks itself by following the instructions. The checks are not a mechanical guarantee and can miss errors, particularly links that would need to be verified against the live web.
The skill is marked as mutating, because saving reports writes files and committing them changes the brain repository. Review what the agent writes and commits, especially when it runs unattended from a cron job.
- No scripts are run by this skill
- No network use is declared
- The link check is performed by the agent and can miss errors
- Saving and committing change files in your brain repository
Tips for getting reliable results
Use the same category names on every run. Inconsistent names make reports hard to find later. If a save returns a write_pending result (exit 10), the link will 404 until the write commits, so poll gbrain write-request with the request ID before sharing the link. If a search comes back empty with a degraded notice, the search was keyword-only, so do not read the empty result as proof that no earlier report exists.
- Keep category names consistent across runs
- Do not link a brain page until it is committed and pushed
- If a link is not available, describe the content instead of linking a profile
- Load only the latest report when that is all you need
What people use it for
- A scheduled cron job saves a morning briefing each day, and you later ask for the latest briefing by keyword.
- An inbox sweep is saved as a dated report so you can compare what the agent found across days.
- A meeting digest is stored in a consistent meeting-sync category and checked for links before it is shared with the team.
- Social mentions are saved and routed by the word "mentions" so you can find recent ones quickly.
Frequently asked questions
Is the reports agent skill free?
The source file reviewed for this page is licensed under MIT. It does not state prices, so check the page where you downloaded the skill for any terms that apply to it.
What do I need to install the reports skill?
You need the skill ZIP, a place to unzip it (~/.claude/skills/ or .claude/skills/ in a project), or the Claude apps with Settings → Capabilities → Skills. To use it fully you also need the gbrain CLI, a git-backed brain repository, and access to skills/_output-rules.md.
How do I install the reports skill in Claude?
Download the ZIP and unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ inside a project. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Does the reports skill run scripts or use the internet?
The skill is listed as running no scripts and using no network. The link checks are carried out by the agent following the instructions, not by a script, so they can miss errors.
Can I use this without gbrain?
The source says the skill is not suited to environments without the gbrain tool or a git-backed brain repo, because the link checks depend on git status and gbrain get commands. Without those, the saving and link checks described here will not work as written.
Why does the link check block some links?
Links to profiles, homepages, channels, unpublished brain pages and placeholders are blocked because they do not take the user to the specific content. Missing links only produce a warning, since a missing link is more honest than a wrong one.
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
This skill suits people running scheduled AI jobs that produce regular summaries, and anyone using gbrain who wants reports stored and searched in a consistent structure. It is a poor fit for one-off chat or writing tasks with no saved reports, or for setups without gbrain and a git-backed brain repo. Because the link check is run by the agent rather than a script, verify important links yourself before you rely on them.
Download the skill