Enhancement: Skill Workflows to Chain Related Skills into Executable Sequences

[Riseonelimit]

Overview
SkillNote currently manages skills as isolated, independent units. There's no way to group related skills into a sequence and run them as a coordinated workflow. This proposal introduces Skill Workflows, a first-class feature that lets you compose multiple skills into an ordered, executable pipeline.

---

Problem

Many real-world agent tasks aren't a single skill, they're a sequence:

"Set up a new feature" =
  1. scaffold-component
  2. write-tests
  3. update-docs
  4. run-linter

Right now a user has to:

• Manually invoke each skill one by one
• Remember which skills belong together and in what order
• Re-do that coordination every time the task comes up
• Hope the agent picks the right sequence on its own

There's no way to encode that multi-step knowledge once and reuse it.

---

Proposed Feature: Skill Workflows

A Workflow is a named, ordered collection of skills with optional conditions, inputs, and branching. It gets exposed to agents as a single MCP tool, the agent calls one thing and SkillNote orchestrates the rest.

---

How It Works End to End

User creates a Workflow in SkillNote Web UI
        ↓
Adds skills as steps, sets order and conditions
        ↓
Workflow saved to DB, exposed as a single MCP tool
        ↓
Agent calls the workflow tool with a task description
        ↓
MCP server resolves each step in order
        ↓
Each skill's content is injected into agent context sequentially
        ↓
Agent executes each step using that skill's instructions
        ↓
Workflow completes  outcome logged to analytics

---

Analytics

• Track workflow-level success/failure rate separately from individual skills
• Show which step in a workflow fails most often
• Per-workflow ratings distinct from per-skill ratings
• A workflow run timeline showing how long each step took

---

Why This Is High Value

Current State	With Workflows
Agent has to figure out multi-step sequences itself	Sequence is encoded once, reused forever
Institutional process knowledge lives in people's heads	Captured as a versioned, shareable workflow
Skills are isolated units with no relationship	Skills are composable building blocks
No visibility into multi-step task performance	Full analytics per workflow and per step

---