Index of every page on this site: llms.txt
All skills
tristanmanchester avatar

/fabric-cli

@3323bc9
by Tristan Manchestertristanmanchester/agent-skills3 stars
2
Run it yourself

Operate the official Fabric.so CLI to search a selected workspace, save notes, links and files, manage tasks, or maintain explicitly authorised project memory. Use for Fabric.so terminal workflows. Not Microsoft Fabric, Python Fabric SSH, Fabric.js, or Daniel Miessler's Fabric framework.

Use this Skill: https://skilld.dev/gh/tristanmanchester/agent-skills/fabric-cli

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈75 tokens for metadata: the name and description. ≈1.2k when used: this file. ≈720 more on demand in 1 file.

Description uses 3.8% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈301 characters, or 3.8%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Fabric.so CLI workflows

Use the native CLI directly or an already authorised Fabric connector when that is the available surface. Do not wrap it in a second command catalogue or treat a planner's read label as an access-control guarantee.

Verify tool and workspace

fabric --version
fabric --help
fabric --json workspace current
fabric --json workspace list
fabric search --help

Verify this is Fabric.so, not another executable with the same name. Help/version checks are local; workspace checks access account state. Check exit status and the returned workspace before any write. A workspace name may be ambiguous; preserve provider IDs where exposed. Switching workspace changes persistent CLI context: serialise operations or use a documented per-command scope, and recheck the target.

The official installer is https://fabric.so/cli/install.sh. Download and inspect it before an authorised installation; do not pipe remote code into a shell by default. Prefer browser login for setup. fabric auth <api-key> puts a secret in process arguments/history; use a supported protected authentication route and never log or copy credentials into notes. Do not invent environment support that the installed CLI has not documented.

Search and retrieve evidence

fabric --json search "project notes"
fabric --json search "design" --tag work
fabric --json path
fabric --json path "Inbox/Reports"

Use search for candidate content and path browsing for known folders. Keep query, tag, workspace, and pagination limitations visible. A snippet is not the full source. For a detailed summary, retrieve original content through a supported CLI/API/connector action. Do not silently substitute the Fabric AI assistant's interpretation for the source document or assume a search is exhaustive.

fabric ask is delegated agent execution. Its documented purpose includes asking the assistant to do things. A natural-language 'read-only' prompt or the old ask-read planner label does not disable tools, writes, or paid inference. Use it only with deliberate scope; use actual resource reads for evidence.

Save the exact content and type

For authorised writes, use note, link, or file when the type is known. save auto-detects type: text resembling a path can upload a file instead of creating a note. Verify that inference before using it.

fabric --json note --parent "Work/Projects" < "/private/output/reviewed-note.md"
fabric --json link "https://example.com" --title "Reference" --parent "Work/Reading"
fabric --json file "/absolute/path/reviewed-report.pdf" --parent "Work/Reports"

The note command's positional text is the body, not a title. Set a separate title only through a flag advertised by the installed command. Inspect the exact bytes and attachment scope before upload, preserve destination constraints, and retain the returned ID. Verify the created resource by ID where supported; search indexing delay can make absence inconclusive.

Tasks and mutations

Read fabric --json task list and resolve the exact task before completion/edit/ delete. Use current task help for fields, priority, due date, and identifiers. Keep due dates distinct from reminder delivery and timezone-bearing times. Preview an explicit ID set for bulk operations; do not rerun a changing query as the plan. Deletion, workspace changes, and remote assistant execution need actual user scope, not permission inferred from a flag or bundled recipe.

Check stdout, stderr, exit status, and the expected response shape separately. Never rerun a mutation without --json merely because parsing failed. It may already have succeeded. Capture the original result, reconcile by ID/current state, and report unknown outcomes. Retry bounded reads when appropriate; no undocumented idempotency is assumed for creates, saves, or delegated asks.

Project memory

Use memory and handoff only when the user or established project instructions authorise persistent Fabric memory. Retrieve a small relevant set, preserve provenance, and draft a compact note before saving it. Do not upload raw conversations or trust a heuristic redactor to remove all secrets.

Finish with evidence

Report the verified workspace, found/changed IDs, relevant source content, and unresolved outcomes. Generated completion scripts are executable shell code; inspect them before sourcing or editing shell startup files. No CLI command in this skill has been authenticated merely because its syntax appears here.

Current official source, reviewed 2026-09-13: Fabric CLI guide. For application SDK work use fabric-api, not these terminal recipes.

Source: SKILL.md on GitHub

Skill behaviors

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

Third-party checks

1 alert1mo3 checks · Risk HIGH
  • Gen Agent Trust Hub1mo

    The skill facilitates management of a Fabric.so workspace but includes high-risk instructions and recommendations to download and execute an installation script from a remote source directly into a shell (curl | sh). While the skill provides safety advisories, such as manual inspection and user consent, this remains a significant security risk. Additionally, multiple bundled scripts use subprocesses to execute commands and scripts.

  • Socket1mo

    No alerts

  • Snyk1mo

    Risk: MEDIUM · 1 issue

Provenance

Signed by skilld at 3323bc9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Steadyupdated last month

Capability

Other metadata
compatibility
Requires the official fabric executable, shell access, and an authorised Fabric.so account for live operations. Installed subcommand help defines the available flags.
metadata
{
  "version": "3.0.0",
  "reviewed": "2026-09-13",
  "source": "https://fabric.so/guide/ai-tools/CLI-usage"
}

README badge

README badge for tristanmanchester/agent-skills/fabric-cli