Skills
Install reusable capabilities for your Super AgentOS. The Skill Store includes 51 maintained verified skills across official packs, plus community-published extensions.
On this page
What is a skill?
A skill is an installable capability your Super AgentOS can use. Skills can be installed, versioned, tracked, and executed through the same platform APIs as built-in tools.
Use skills when you want to add focused domain logic without creating a full new agent. Good examples are CSV processing, prompt evaluation, ticket prioritization, release-note generation, or PII redaction.
Browser session vs bearer token
The web app uses a secure browser session by default. That means installs, store actions, and Studio commands work after you sign in once without pasting a token into the UI.
Generate a bearer token only when you need to call AgentOS from an SDK, another machine, automation, or a third-party integration.
Official verified skill packs
AgentOS maintains official verified skills in packs so you can install a coherent set of tools quickly. Each skill below has a detail page and can be installed directly.
Core Utilities
Baseline transformation, formatting, and request-building helpers for most agents.
Normalize text, casing, slugs, and simple string transforms.
Compute averages, sums, and small analytical rollups.
Parse CSV text and aggregate selected columns quickly.
Extract and reshape nested JSON payloads.
Build headers, query strings, and response helpers for HTTP workflows.
Format dates, add durations, and compare time windows.
Extract, test, and replace text using regular-expression helpers.
Generate checklists, headings, and markdown sections from data.
Deduplicate, chunk, sort, and summarize arrays.
Parse origins, query strings, and normalized URL components.
Render small tokenized templates for messages and reports.
Validate required fields and common input constraints.
Finance and Ops
Portfolio, risk, and operational calculation helpers for finance-heavy workflows.
Calculate margins, fees, and simple financial ratios.
Compare target allocations against current portfolio weights.
Score exposures using simple weighted rules.
Inspect OHLC candles for body, wick, and direction metrics.
Research and Support
Knowledge, summarization, ticket routing, and SLA helpers for research and support teams.
Structure research notes, highlights, and evidence summaries.
Format citations consistently for research outputs.
Track experiment variants, statuses, and small result summaries.
Summarize survey answers and compute quick response metrics.
Split long text into retrieval-friendly sections.
Measure response windows and SLA breach risk.
Rank support tickets by urgency and business impact.
Classify basic sentiment with deterministic rules.
Turn chat transcripts into compact summaries and action items.
Route cases to the right team based on rules and urgency.
AI and Evals
Prompt prep, evaluation, labeling, and heuristic classification utilities for AI workflows.
Extract variables, generate prompt variants, and inspect prompt structure.
Calculate pass rates and weighted evaluation summaries.
Label common workflow items with deterministic classifier rules.
Estimate tokens and prepare content for embedding pipelines.
Count labels and enforce dataset annotation consistency.
Communication and Delivery
Message, outreach, meeting, and release communication helpers for shipping teams.
Generate subject lines, greetings, and structured outbound email drafts.
Compose concise status, alert, and handoff message blocks.
Turn raw notes into decisions, owners, and next-step summaries.
Generate follow-up sequences and cadence suggestions.
Normalize names, phone fields, and role labels for contact lists.
Check required environment variables for missing or empty values.
Generate deploy readiness checklists and release gates.
Convert change items into customer-facing release notes.
Prepare incident updates and maintenance window notices.
Structure timelines, causes, and follow-up action items after incidents.
Security and Analytics
Governance, redaction, anomaly, KPI, and forecasting helpers for safer production ops.
Evaluate passwords against baseline policy rules.
Mask email addresses, phone numbers, and common identifiers in text.
Normalize audit events into consistent report-friendly records.
Flag obvious token and credential patterns in text blobs.
Summarize role grants and review overdue access records.
Build safe where-clause fragments and query summaries.
Convert raw KPI values into scored status summaries.
Calculate retention and grouped cohort metrics from labeled rows.
Flag spikes and drops with threshold-based anomaly rules.
Project simple forward trends using lightweight averages.
Installing skills
In the web app, open the Skill Store and install directly while signed in. For external clients, call the install endpoint with a bearer token.
POST /api/skills/install
Authorization: Bearer <your-bearer-token>
Content-Type: application/json
{ "skill_id": "uuid-of-the-skill" }
// -> { "success": true, "installation": { "id": "...", "installed_at": "..." } }Using installed skills
Once a skill is installed, run one of its capabilities through POST /api/skills/use.
POST /api/skills/use
Authorization: Bearer <your-bearer-token>
Content-Type: application/json
{
"skill_slug": "json-transformer",
"capability": "extract",
"params": {
"object": { "version": "v2", "region": "lagos" },
"path": ["version"]
}
}
// -> { "success": true, "result": "v2", "execution_time_ms": 4 }Building your own skill
A skill source file defines a class named Skill. Each public method becomes a callable capability.
class Skill {
summarize(params) {
const text = String(params.text || '');
const maxLength = Number(params.maxLength || 120);
return text.length <= maxLength ? text : text.slice(0, maxLength) + '...';
}
}- Return JSON-serializable values only.
- Declare capabilities clearly so Studio and Skill Store detail pages can show them.
- Ask for only the primitives you really need.
- Keep methods deterministic unless side effects are the explicit purpose of the skill.
Publishing to the Skill Store
Use the Developer Dashboard while signed in, or call the API directly from an external client with a bearer token.
POST /api/skills
Authorization: Bearer <your-bearer-token>
Content-Type: application/json
{
"name": "My Skill",
"slug": "my-skill",
"category": "Utilities",
"description": "One sentence summary",
"capabilities": [{ "name": "run", "description": "Runs the skill", "params": { "input": "string" }, "returns": "string" }],
"source_code": "class Skill { run(params) { return String(params.input || '') } }"
}