Skip to content
Skills Library
AI & agents · Automation

cortex-skills-loop

/cortex-skills-loop

Drives the cortex skills recommend-feedback-rate loop. Use when a context change occurs (new file types, domain shift, task pivot) or when a task completes and skill effectiveness should be recorded.

Licence
MIT
Last repository change
28.06.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 28.06.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Drives the cortex skills recommend-feedback-rate loop. Use when a context change occurs (new file types, domain shift, task pivot) or when a task completes and skill effectiveness should be recorded.

Original author description

Drives the cortex skills recommend-feedback-rate loop. Use when a context change occurs (new file types, domain shift, task pivot) or when a task completes and skill effectiveness should be recorded.

How it works

  1. Overview
  2. When to Engage
  3. Workflow
  4. Closing the Loop
  5. Quick Reference
Original SKILL.md headings · 125 lines, 718 words
SKILL.md preview
---
name: cortex-skills-loop
description: Drives the cortex skills recommend-feedback-rate loop. Use when a context change occurs (new file types, domain shift, task pivot) or when a task completes and skill effectiveness should be recorded.
keywords:
  - skills recommend
  - skill feedback
  - context change
  - cortex
  - skills
  - loop
  - cortex skills loop
---

# Cortex Skills Loop

## Overview

The cortex CLI includes an AI-powered recommendation engine that learns from usage patterns. This
skill establishes the workflow for participating in that learning loop: get recommendations when
context shifts, provide feedback on recommendation quality, and rate skills after use. Each
interaction improves future recommendations.

## When to Engage

### Signals That Trigger Recommendations

Run `cortex skills recommend` when any of these mismatch signals appear:

- **File pattern shift** — `git diff` or the working set includes file types not covered by
  active skills (e.g., `.tf` Terraform files appear but no infrastructure skill is active,
  or `**/auth/**` paths get touched without security skills loaded)
- **Agent activation change** — a new agent gets activated that likely has complementary skills
  not yet loaded (the recommender maps agents to skill sets internally)
- **Explicit domain pivot** — the user switches focus to a different domain ("now let's handle
  the database migrations") and the current skill set is oriented elsewhere
- **Skill gap felt** — a task requires domain knowledge that no active skill covers, or an
  active skill is providing no value to the current work

### Signals That Trigger Rating/Feedback

- **Task completion** — a skill was active during work that just finished successfully
- **Recommendation acted on** — a recommendation was recently followed or dismissed
- **Negative experience** — a skill actively misled or produced unhelpful guidance

### When Not to Engage

Do **not** run recommendations on every session start or for routine tasks where the active
skill set is clearly appropriate. The loop adds value through selective, signal-driven use.

## Workflow

### 1. Recommend — Surface Relevant Skills

When a context change is detected, run:

```bash
cortex skills recommend
```

This analyzes the current git state, active agents, and historical patterns to suggest skills
grouped by confidence level:

- **High confidence (>=0.8):** Auto-activate candidates — consider activating immediately
- **Medium confidence (0.6-0.8):** Worth reviewing with the user
- **Low confidence (<0.6):** Informational only

To check what is currently active before acting on recommendations:

```bash
cortex status
```

For full option details: `cortex skills recommend --help`

### 2. Feedback — Improve Recommendation Quality

After a recommendation has been acted on (activated or dismissed), record whether it was useful:

```bash
cortex skills feedback <skill-name> helpful --comment "Caught auth vulnerability early"
cortex skills feedback <skill-name> not-helpful --comment "Not relevant to this API work"
```

Positive feedback with context available teaches the recommender to associate the current
project context with that skill for future sessions. Always provide a `--comment` when possible
to enrich the learning signal.

For full option details: `cortex skills feedback --help`

### 3. Rate — Record Skill Effectiveness

After a task completes and a skill was active during the work, rate its contribution:

```bash
cortex skills rate <skill-name> --stars <1-5> --review "Description of experience"
```

Additional flags to enrich the signal:

- `--helpful` / `--not-helpful` — binary usefulness indicator
- `--succeeded` / `--failed` — whether the task the skill supported succeeded

Ratings feed back into the recommendation engine. Highly rated skills get prioritized in
future suggestions; low-rated skills get demoted.

To view existing ratings before adding one: `cortex skills ratings <skill-name>`

To discover top-performing skills: `cortex skills top-rated`

For full option details: `cortex skills rate --help`

## Closing the Loop

The recommend-feedback-rate cycle is cumulative. Each interaction updates the SQLite-backed
learning database, improving four recommendation strategies:

1. **Semantic similarity** — embeds successful session contexts for future matching
2. **Rule-based** — matches file patterns to skill suggestions
3. **Agent-based** — maps active agents to complementary skills
4. **Pattern-based** — promotes skills with high historical success rates

When multiple strategies converge on the same skill, confidence gets boosted. Consistent
feedback and ratings are what make this convergence happen over time.

## Quick Reference

| Action | Command |
|--------|---------|
| Get recommendations | `cortex skills recommend` |
| Check current status | `cortex status` |
| Give positive feedback | `cortex skills feedback <skill> helpful --comment "..."` |
| Give negative feedback | `cortex skills feedback <skill> not-helpful --comment "..."` |
| Rate a skill (1-5 stars) | `cortex skills rate <skill> --stars N --review "..."` |
| View skill ratings | `cortex skills ratings <skill>` |
| See top-rated skills | `cortex skills top-rated` |
| View usage analytics | `cortex skills analytics` |

Versions

The database has no other skills named cortex-skills-loop.

Identical content appears in 1 other GitHub folders. We show one copy with the highest score.

Database snapshot: 13.11.2026 · package version 30dc96a22612 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100cortex-run
    ONLY load this skill when the user explicitly types $cortex-run or /cortex-run. NEVER load this skill from auto-routing hooks or keyword matching. For auto-routed prompts, use snowflake-cortex-code:cortex-router instead.
  • 100cortex-setup-project
    Bootstrap Cortex for a new project or import existing session history. Use when the user says 'set up Cortex', 'seed this project', 'import my history', 'backfill memories', 'bootstrap memory', 'initialize Cortex for this project', or when starting to use Cortex on an existing…
  • 100cortex-evolve
    Save knowledge to the cortex vault. Use when the user says "存到 cortex", "記一下", "save to cortex", "記到筆記", "cortex evolve", or when the session-suggest hook recommends saving content. Handles Notes and Projects.
  • 96cortex-mcp-project
    Use when working on the Cortex MCP server repo.
  • 90using-cortex
    Use when the user asks to check the cortex vault, refers to earlier work ("last time", "that project", "we discussed", "之前那個", "上次的"), asks to resume something from a previous session, or names a topic the SessionStart hook listed for this vault. Routes the request to the…
  • 100cortex-m
    Build, test, or develop the Cortex-M (CMSIS-NN) backend. Use when working on backends/cortex_m/, running Cortex-M tests, or exporting models for Cortex-M targets.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.