Build repeatable AgentWorks browser validation for authentication and session behavior. Use for login, logout, MFA, recovery, expiry, refresh, and invalid-session coverage after browser setup.
Installs into .claude/skills of the current project.
Are you the author of Authentication Session Validation?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/manishiitg-authentication-session-validation)
---
name: authentication-session-validation
description: Build repeatable AgentWorks browser validation for authentication and session behavior. Use for login, logout, MFA, recovery, expiry, refresh, and invalid-session coverage after browser setup.
---
# Authentication and Session Validation
## Outcome
Create repeatable, role-aware tests for the application's approved authentication and session lifecycle with durable results and attempt-scoped evidence.
## When to use
Use after Basic Browser Setup when authentication behavior needs coverage beyond establishing one working test session. Include only flows enabled and authorized for the customer environment.
## Discovery and user direction
Inspect the current workflow, goals, metrics, configuration, capabilities, stores, reports, and triggers before proposing changes. Summarize reusable design and gaps, then ask focused questions for unresolved customer choices such as scope, success, approvals, thresholds, ownership, and budgets. Record the answers as customer direction. Installation alone does not approve workflow changes or execution. Default to one small-team workflow; split only for incompatible access or lifecycle boundaries.
## Required inputs
Resolve authentication methods, actor roles, approved expectations, account ownership, MFA/recovery channels, session lifetime rules, lockout limits, allowed state changes, evidence retention, and cleanup/reset procedures.
## Plan and AgentWorks tools
Use scripted steps for repeatable Playwright scenarios and deterministic persistence. Use a message sequence for evidence-based investigation. Add branches only for real outcome paths or configured review. Reuse selected secrets and never pass secret values through plan context.
## Knowledge and persistence
Keep secret references in workflow configuration. Store verified routes, session behavior, role requirements, and non-secret auth quirks in application KB notes. Store runs, cases, attempts, and artifact metadata in the DB and video plus redacted console/network evidence in durable assets.
## Validation and reporting
Require a terminal result for every expected auth scenario and actor. Verify fresh-context behavior, post-action state, required cleanup, source/build identity, and evidence status. The dashboard shows the scenario/actor matrix, environment and build, failure class, evidence status, and history without exposing credentials or tokens.
## Guardrails
Do not trigger unapproved lockouts, recovery messages, enrollment changes, account deletion, or production authentication experiments. Never persist passwords, cookies, tokens, MFA seeds/codes, storage state, or unrestricted network bodies.
## Read details when needed
- [Workflow design and outcomes](../../references/workflow-design-and-outcomes.md): goals, metrics, and current-versus-separate workflow decisions.
- [AgentWorks plan and tools](../references/agentworks-plan-and-tools.md): step and platform-tool choices.
- [Evidence capture](../references/evidence-capture.md): durable, redacted attempt evidence.
- [Authentication guide](references/authentication-workflow.md): scope, flow, storage, and acceptance cases.
- [Example scenarios](examples/authentication-scenarios.json): fictional coverage shape.
- [Catalog metadata](playbook.json): presentation and optional recommendations.
## Completion contract
Return installed playbook and scenario revisions, customer overrides, actor/scenario coverage, actual trial results, KB/test/evidence/report locations, capability resolution, and unresolved or unsafe-to-test behavior.