> ## Documentation Index > Fetch the complete documentation index at: https://docs.cantina.security/llms.txt > Use this file to discover all available pages before exploring further. # Skills > Reusable playbooks that teach agents how to handle specific classes of tasks to a known standard. A skill is a reusable playbook that teaches the agent how to perform a specific class of tasks to a known standard, along with the tools available to investigate and remediate security issues. Agents use skills to make decisions during triage. Updating a skill is how you change agent behavior. ## What a skill contains * **Content**: Rich procedural text describing the playbook, with headings, lists, links, and embedded tools. * **Tools**: Integrations the skill can invoke during execution, including Slack, Teams, Discord, SMS, email, PagerDuty, webhook, DNS, Jira, AWS, Vercel, Huntress, CrowdStrike, plus custom MCP tools. * **Domain tags**: Categories like DNS, Identity & Access, Web3, Cloud, SDLC. * **Versions**: Every save creates a snapshot with editor attribution. Compare and roll back as needed. * **Lineage**: Skills can be **managed** (synced from a template) or **unmanaged** (forked or custom). Editing a managed skill forks it. ## Skill templates Clarion ships **131 skill templates** across **13 domains** — Cloud, Identity & Access, Endpoint & Device, SDLC, DNS & Infrastructure, Operations, Vulnerability, Security Incidents, Network, Web3 & Blockchain, Application, Email, and Threat Intelligence. See [Our skills](/clarion/learn/skills#our-skills) for the full breakdown, and the [Community Library](https://clarion.cantina.xyz/community) for skills shared by other teams running Clarion. ## Agent behavior Skills describe how to handle a specific class of tasks. For guidance that should apply across **every** agent in your workspace, use [**Settings → Agents → Agent behavior**](/clarion/learn/settings#agent-behavior) instead of duplicating it in each skill. Good examples: * How to start a war room. * Organization-wide terminology, escalation hierarchy, and on-call conventions. ## What you can do * **Create** skills from a template or from scratch. * **Edit** skill content with full version history. * **Embed tools** in a skill to give agents the actions they need. * **Assign** skills to agents. * **Fork** templates on customization, or **reset** back to the template default. Every domain we ship skill templates for. The actions a skill can give an agent. How to write a skill agents pick at the right moment and follow well. ## Our skills Clarion ships **131 skill templates** across **13 domains**. Every skill is a self-contained playbook: its description says when to apply it, and its body says how to investigate and what to do about it. Attach one to an agent and it stays synced with the template; edit it and it forks into your own copy. Skills are grouped by domain so you can find the right playbook by the kind of problem, not by which vendor raised it. A skill can be shared by several agents — the counts below are of distinct skills. | Domain | Skills | Coverage in one line | | - | - | - | | [Cloud](#cloud) | 40 | AWS, Azure, GCP, Cloudflare, Vercel, Replit, and Wiz control-plane and exposure response | | [Identity & Access](#identity-access) | 26 | Sign-in anomalies, MFA and session attacks, OAuth abuse, privilege escalation, secret-manager misuse | | [Endpoint & Device](#endpoint-device) | 16 | EDR triage and the full attack chain, from execution to ransomware, across four vendors | | [SDLC](#sdlc) | 13 | Repository and CI/CD integrity, dependency findings, MCP-server governance, shift-left design review | | [DNS & Infrastructure](#dns-infrastructure) | 10 | Record, subdomain, and email-auth changes; registrar hijack, tunneling, and takeover risk | | [Operations](#operations) | 7 | Outage runbooks, detection tuning, knowledge upkeep, vendor review, questionnaire drafting | | [Vulnerability](#vulnerability) | 6 | Dependency and bug-bounty findings, reachability validation, remediation evidence | | [Security Incidents](#security-incidents) | 5 | Cross-product incident entry points and hunting pivots | | [Network](#network) | 3 | Edge DDoS, WAF, and origin-exposure response | | [Web3 & Blockchain](#web3-blockchain) | 2 | On-chain alert response and multisig signer monitoring | | [Application](#application) | 1 | Client-side script and supply-chain integrity | | [Email](#email) | 1 | Phishing and email-threat investigation | | [Threat Intelligence](#threat-intelligence) | 1 | Advisory triage and applicability to your environment | ### Cloud **40 skills.** The largest domain, and the one with the deepest per-provider coverage. * **AWS (10)** — a general CloudTrail response procedure plus dedicated playbooks for root-account security, IAM privilege escalation and persistence, credential exposure and access-key compromise, console authentication abuse, compute compromise, network and infrastructure manipulation, data exfiltration and resource exposure, suspicious access activity, and defense evasion or log tampering. * **Azure (10)** — security posture triage and Defender for Cloud alert response, then Entra ID identity compromise, RBAC privilege escalation and persistence, Storage and data exposure, Key Vault secret access, network exposure and lateral movement, compute compromise, defense evasion, and Secure Score misconfiguration remediation. * **GCP (7)** — Cloud Monitoring incident triage and Security Command Center finding triage, then privilege escalation and anti-forensics, service-account key compromise, unauthorized compute, and SCC-specific IAM and exposure response. * **Wiz (3)** — toxic combinations scored on whether their ANDed facets are simultaneously live, threat detections scored on actor and baseline, and resource and identity ownership context. * **Cloudflare (2)** — origin and edge availability, certificate and Access lifecycle. * **Vercel and Replit (6)** — exposure sweeps and single-site triage for each platform, plus Vercel deployment-failure triage and security and domain monitoring. * **Microsoft Sentinel (1)** — Azure and cloud control-plane activity investigation. * **Cross-provider (1)** — suspicious shell session. ### Identity & Access **26 skills.** The broadest attack-pattern coverage of any domain, mirrored across identity providers. * **Provider-neutral core (14)** — triage and a general response procedure, then impossible travel, brute force and credential attacks, MFA attacks and session hijacking, OAuth and API abuse, admin compromise and privilege escalation, federation and IdP abuse, email and communication exposure, data exfiltration and staging, mobile device and MDM security, security control and policy weakening, government-backed attack warnings, and phishing campaign response. * **Microsoft (6)** — Defender XDR variants of impossible travel, MFA and session hijacking, OAuth abuse, and admin compromise; Sentinel identity-compromise investigation and UEBA anomaly investigation. * **1Password (4)** — burst failed sign-ins, user suspension, vault item export, and the vault-access-then-export sequence. * **Huntress ITDR (1)** and **Wiz entitlement review (1)** — identity findings from the detection and posture sides. ### Endpoint & Device **16 skills.** Every vendor gets a triage entry point; the attack-chain playbooks sit behind it. * **CrowdStrike (8)** — Falcon and OverWatch triage, then execution (scripts, LOLBins, process-ancestry anomalies), persistence, credential access, defense evasion and sensor tampering, lateral movement, command and control and exfiltration, and ransomware and destructive action. * **SentinelOne (4)** — threat triage, investigation and scoping, gated containment and response, and false-positive handling with exclusions. * **Microsoft (2)** — Defender endpoint host deep-dive and Sentinel ransomware and mass-impact scoping. * **Huntress (2)** — incident investigation and escalation handoff. ### SDLC **13 skills.** Repository integrity, pipeline integrity, and the governance of what developers connect their tooling to. * **GitHub (7)** — tag spraying, CI/CD script injection via branch names, force push on a default branch, repository visibility change to public, branch-protection misconfiguration, org-owner addition, and a generic GitHub alert fallback. * **Claude Code / MCP governance (3)** — server governance triage, inventory review sweep, and external account triage. * **Apex (2)** — triage and remediation, plus a customizable Apex skill. * **Shift-left (1)** — assessing new Jira epics for security risk before implementation starts. ### DNS & Infrastructure **10 skills.** Change detection on the records that define who you are, plus takeover risk. Record change, subdomain change, email-authentication (SPF/DKIM/DMARC) change, new subdomain discovery, DNS hijacking and registrar compromise, DNS tunneling and data exfiltration, DNS DDoS and amplification, Cloudflare DNS and zone integrity, and the dangling-subdomain sweep and triage pair. ### Operations **7 skills.** The playbooks that keep the platform and the program healthy rather than responding to an attack. Outage runbook, health-repair triage for Clarion's own integrations and monitors, signal-rule refinement, Brain maintenance, Vercel production rollback and recovery, review of newly discovered Vanta vendors, and drafting answers to vendor security questionnaires (DDQ / SIG / VSA). ### Vulnerability **6 skills.** Findings that arrive as a queue rather than an incident, and the evidence trail behind closing them. Dependabot vulnerability remediation and backlog processing, reachability and runtime validation with exposure scoring (shared by Apex and Dependabot), Wiz vulnerability and component exposure, Cantina bug-bounty finding triage, and SOC 2 evidence collection. ### Security Incidents **5 skills.** Cross-product entry points that route to the scenario playbooks in other domains. Microsoft Defender XDR and Microsoft Sentinel incident triage procedures, the Sentinel advanced-hunting pivot, Cloudflare Logpush incident handoff, and credential-harvesting triage. ### Network **3 skills.** Cloudflare Logpush L7 DDoS detection triage, Logpush origin-exposure and DNS-bypass checks, and Cloudflare DDoS and WAF attack response. ### Web3 & Blockchain **2 skills.** Smart contract and on-chain alert response — including emergency pause coordination — and multisig signer monitoring. ### Application **1 skill.** Cloudflare client-side security response, covering injected JavaScript and browser-side supply-chain integrity. ### Email **1 skill.** Microsoft Sentinel phishing and email-threat investigation. ### Threat Intelligence **1 skill.** Threat intel advisory triage and applicability — deciding whether a published advisory, IOC set, or TTP actually touches your environment. ### Beyond the library Skill templates are a starting point. The **[Community Library](https://clarion.cantina.xyz/community)** carries skills written and proven by other security teams running Clarion, alongside full agents. Take a single skill into an agent you already have, or an entire agent with its skills and scheduled jobs. Skills and agents shared by teams running Clarion. Every agent template and the skills it bundles. ## Authoring best practices Skills are the backbone of our platform. Authoring good skills will make a huge difference in both the triage result, and also the duration. These practices follow the skill templates we ship as part of Clarion. Open any of them from the [skill library](/clarion/learn/skills#our-skills) to see an example. ### What is a skill A skill is a set of instructions written in natural language that belong to one or multiple agents. They are primarily used as part of the triage and remediation phases, thus authoring your own skills gives you control over how different alerts are being handled. ### How agents use skills Every agent has access to its own skills plus any skills referenced by those. Also, it has access to any skills referenced in [agent behavior](/clarion/learn/settings#agent-behavior) or in the scheduled task prompt. When starting, the agent gets the title and description of each of those skills. It will then decide which ones to follow, and read their full body. After that, it will proceed to follow the instructions in those skills it read. We don't inject the full skill body in the agent context, to prevent bloat. That's why good titles and descriptions are essential to make sure the agent picks the right one. ### Title A good title should name the category of this task, not the agent or author. As a reader scanning the catalog, we should know what the skill is for from the title alone. | Good | Weak | | - | - | | Identity Entitlement Review | Wiz skill 2 | | Vercel Exposure Sweep | Weekly check | | Credential harvesting triage | Claude Code agent playbook | Include the vendor when the skill only works against one vendor's data. Keep it generic if it's provider neutral ### Description The description will be used by the Agent when to decide what skill to follow, so it must include two things: * a short summary of what the skill does * instructions of when to use this skill Keep the description short, ideally 2 or 3 sentences. A long description bloats the context of every agent run Good description examples: * Investigates GCP service account key creation as potential credential compromise or attacker persistence. Use when a new service account key is created, especially outside expected workflows. * Investigates failed-sign-in bursts on a 1Password user as brute force, credential stuffing, or targeted account takeover. Use when an alert fires for multiple failed 1Password sign-in attempts for the same user in a small period. * Remediates Dependabot vulnerabilities end-to-end and escalates high/critical findings in place. Use when a Clarion alert fires for one or more Dependabot vulnerabilities in a repository. Bad description examples: * Handles Wiz findings. * Investigates suspicious activity and takes the appropriate action. * First fetch the issue, then query the audit logs for the last 24 hours, check the user in Okta, and post a summary to Slack. ### Body The body represents the instructions for the agent to follow. Try authoring it like guidance from the most experienced person on the team, thus saying what matters, what to check, what the evidence means, and what to do about it. Agents are smart so this doesn't need to be a very specific step-by-step list. Keep it generic when the flow is not clear, so the agent can follow its own reasoning #### Common template Most templates follow the same outline. Keep the sections your skill needs and drop the rest. | Section | Purpose | | - | - | | Opening framing | One to three sentences on the goal and on what does and does not count as a finding | | Ground rules | Rules that hold across every step, each with its reason | | Workflow | Usually numbered steps, each naming what to check, and how/if the result changes the next step | | Outcome | The evidence that decides the verdict and severity, then what to do once decided: escalate or set the status, and write what was learned to the Brain | | Remediation | What the agent may fix itself and under which verdict (open a PR, revert a record, contain a host) and what it leaves to a human as a follow-up task | #### What to put in the skill 1. Specify what the goal is and what isn't > The goal is to find deployments open to the internet that were **not meant to be**: internal dashboards, admin panels, stale proofs-of-concept. Intentionally public marketing sites, blogs, and products that enforce their own login are not findings. 2. Give every rule its reason so the agent can generalise > Do not escalate Dependabot issues that are related to dev-only packages. They create noise and unnecessary work for the team. The priority should be on actual packages used in the production deployment. 3. Mention the required evidence Say what counts as actual proof and what does not. The correctness of the alert details cannot be guaranteed, that's why the agent needs to correlate with actual evidence by using the available tools. 4. Keep it concise and clear Think whether or not this skill makes sense if read by a colleague. If not, probably won't make sense if read by an agent. Split a skill when its triggers differ, or when the evidence that decides the verdict differs. Two skills that each do one job route better, and read better, than one skill with a branch at the top. #### What not to put in the skill 1. Do not mention tools by name. They are internal and may change freely, thus your skills will be outdated. Instead specify what the agent should do, not how. 2. Do not mention tool categories. By default each agent has access to all tools connected in your workspace. You can configure this from the agent page if you want granular control. 3. Do not give instructions about where or when to post notifications. Configure both under **Settings → Agents → Escalation & notifications**: see [Notifications](/clarion/learn/notifications) and [Severities that escalate](/clarion/learn/settings#severities-that-escalate). 4. Do not give instructions regarding asking for approvals. Those are set up once under **Settings → Tool policies** and they apply to all agents. 5. Do not put inside skills general knowledge about your business, the systems you use or the clients you have. This knowledge belongs in the Clarion Brain not in skills. 6. Do not put inside skills instructions that apply across all agents. Those belong in [agent behavior](/clarion/learn/settings#agent-behavior). A good example of such a rule is always creating a Linear task before starting triage of an issue. ### References inside a skill The skill editor turns references into pills. Each kind behaves differently at runtime, and two of them change what the agent is allowed to do. | Pill | Editor trigger | What it does at runtime | | - | - | - | | Skill | `$` | Adds the referenced skill to the agent's catalog, even when that skill is not linked to the agent, so the agent can fetch it at the step that needs it. References are followed up to three levels deep. | | Knowledge | `$` | Declares a Brain knowledge element the skill depends on. The agent is told to fetch its live content by id instead of searching for it. The `$` picker lists skills and knowledge together, and offers knowledge only in the skill editor. | | Channel or person | `#`, `@` | Points the agent at a Slack or Teams channel or person. Although notification settings are set up once per workspace, this can be useful to point the agent at channels to read or people to coordinate with during the investigation. | | Tool | `/` | DEPRECATED. Names the tool the step should use. It is no longer needed, as the agent can use every connected integration its agent settings allow. | Edit the skill to change how an agent handles an issue. The agent's own settings decide which tool groups it may use and which model it runs on, so check those when an agent cannot take an action its skill asks for. ### Final recommendations Always review the skills you write, even if written by an AI. Good skills are essential for a good Clarion experience. Before saving them, ask Clarion to review them, it will provide actionable feedback. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.