ramp-complete-tasks

Inspect the Ramp attention feed and route outstanding tasks to their owning workflows.

2|1|Updated Jul 24, 2026
One-click install
npx skills add https://github.com/ramp-public/skills --skill ramp-complete-tasks-ramp-public
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ramp-complete-tasks
Source: https://github.com/ramp-public/skills/tree/main/plugins/ramp/skills/ramp-complete-tasks
Command: npx skills add https://github.com/ramp-public/skills --skill ramp-complete-tasks-ramp-public

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Ramp users accumulate outstanding tasks (missing receipts, revision requests, approvals) that can lock their cards, and it is hard to know what needs attention or how to clear each item. This Skill reads the live attention feed, dispatches each task type to the correct workflow, and verifies that tasks and card locks are actually cleared. ## Core Features & Use Cases - Attention Feed Inspection: Reads the full Ramp attention feed via the GetAttentionFeed MCP tool or ramp tasks list CLI, paginating every section cursor and reconciling counts against total_count. - Task-Type Dispatch: Routes task types such as TRANSACTION_REVISION_REQUESTED, TRANSACTION_MISSING_ITEMS, approvals, procurement, vendor, and card activation tasks to the appropriate owning skill. - Card Sanction Handling: Diagnoses locked cards using is_spendable, active_locks, and blocking_task_count instead of relying on lifecycle state, and resolves task-based locks through the feed. - Postcondition Verification: Polls the feed and ListCards with bounded retries to confirm tasks disappear and cards report is_spendable: true before claiming success. - Use Case: A user asks why their Ramp card is locked. The Skill lists the card, finds a task sanction, reads the attention feed, completes the outstanding expense revision through the owning workflow, and confirms the card is spendable again. ## Quick Start Ask the agent to show what needs your attention in Ramp and complete your outstanding tasks.

Frequently Asked Questions about ramp-complete-tasks

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I see what needs my attention in Ramp?

Use the GetAttentionFeed MCP tool or run ramp tasks list with a rationale to read the live attention feed. The Skill paginates every section cursor until next_cursor is null and reconciles item counts against each section's total_count before summarizing tasks.

Why is my Ramp card locked and how do I unlock it?

A card can be locked by outstanding tasks even when its lifecycle state is ACTIVE. The Skill calls ListCards, inspects is_spendable, active_locks, and blocking_task_count, then resolves task sanctions through the attention feed rather than calling a manual unlock.

How do I complete a transaction revision request in Ramp?

Transaction revision requests are routed to the ramp-complete-expenses workflow, which corrects the requested fields and calls CompleteTransactionRevision or ramp transactions complete-revision with a user-confirmed reason. The task is only reported done after it disappears from the feed.

Does this work with both the Ramp MCP connector and the ramp CLI?

Yes, the Skill supports both surfaces. Every operation has an MCP tool equivalent such as GetAttentionFeed and ListCards, and a matching CLI command such as ramp tasks list, each requiring a non-empty rationale.

What happens when a task write succeeds but the task still appears?

The task is treated as still outstanding. The Skill re-reads its live context, reports what remains, and never claims completion from a successful write call alone; verification requires the task to be absent from the fully paginated feed.

What are the limitations when handling unknown Ramp task types?

Unknown or unsupported task types are never mutated or dismissed. The Skill presents the task context and the returned Ramp link and asks the user to complete it in Ramp, avoiding guessed mutations.