alivirgo/major-ai-skills/vault
Claude Code and agent skill by alivirgo/major-ai-skills
- community
Checking that the SKILL.md file is still in the repository.
Configure Vault auth methods, least-privilege policies, and dynamic secrets with correct K8s TokenReview wiring; never log secret material or treat root tokens as routine.
Install
npx vskill@latest install alivirgo/major-ai-skills/vaultSKILL.md
HashiCorp Vault Secrets AI Skill Guide (Claude)
Overview & Engine Architecture
Vault centralizes secrets, encryption (Transit), and dynamic credentials (DB, PKI, cloud) behind authentication and ACL policies. Clients authenticate → receive a token with lease TTL → access only allowed paths. HA depends on storage backend and unseal procedures (Shamir or auto-unseal)—agents document patterns but never invent production unseal keys.
When to use / when not to
Use when
- Designing KV v2 path layouts, policies, AppRole for CI, Kubernetes auth for pods.
- Issuing dynamic DB credentials vs long-lived passwords in Git.
- Incident response: revoke leases/tokens after leak.
Do not use when
- Storing TLS certs for public websites only →
@lets-encryptmay suffice; Vault PKI for internal mesh. - Replacing full IAM for humans—Vault complements, rarely replaces cloud SSO for operators.
Operational Capabilities & Agent Directives
- Never paste root tokens, unseal keys, wrapped tokens, or secret payloads into chat/logs.
- KV v2 paths: use
secret/data/...for reads/writes;secret/metadata/...for policies; enable versioning; usecasfor concurrent writers. - Least privilege policies: narrow path prefixes; separate read vs admin; test with non-admin token.
- Dynamic secrets preferred over static KV for databases; always revoke leases on rotation/incident (
vault lease revoke -prefix). - Kubernetes auth gotcha: Vault calls TokenReview API—reviewer JWT needs
system:auth-delegator; short-lived reviewer tokens expiring breaks all logins. - Vault in-cluster: omit
token_reviewer_jwt, use local projected SA token with periodic reload (Vault ≥1.9.3); setdisable_local_ca_jwt=truewhen using client JWT as reviewer per docs. - AppRole: short
token_ttl, bindsecret_iddelivery out-of-band; rotate role IDs on compromise. - Audit devices early; protect audit sink integrity.
Production Example: policy + K8s auth role
Policy api-prod.hcl:
Enable and configure (sketch):
ClusterRoleBinding for reviewer SA (when external Vault):
AppRole for CI (short TTL):
Technical Troubleshooting Matrix
| Signature | Likely cause | Fix | |
|---|---|---|---|
403 tokenreviews forbidden | Reviewer SA lacks delegator | ClusterRoleBinding system:auth-delegator | |
| All K8s logins fail after ~24h | Expired reviewer JWT | In-cluster Vault or rotate long-lived reviewer | |
permission denied on path | Policy typo / wrong mount | vault token capabilities with test token | |
| Leaked approle secret_id | Credential theft | Rotate secret_id; revoke accessors | |
| Sealed vault | Restart/unseal event | Operator unseal ceremony—agents don’t guess keys | |
| DB creds still work after revoke | App cached password | Revoke lease + restart app |
Best Practices
- Namespaces (Enterprise) or strict path prefixes for team isolation in OSS.
- Break-glass root procedure documented offline; no root in Git.
- Prefer Transit for app-level crypto without exporting key material.
- Pair with
@kubernetesCSI/driver or external-secrets for pod injection. - Run rotation drills; measure time-to-revoke.
Limitations
- Vault outage is critical path—design HA, backups, and unseal custody.
- Mis-scoped policies are common; always validate with dedicated test tokens.
- Agents must stop and ask for seal status, namespace, and auth method before destructive revoke.
Related Skills
@kubernetes— service accounts, projected tokens, injectors@terraform— Vault provider resources (if used)@github-actions— AppRole/OIDC for CI secrets@mysql/@postgresql— dynamic DB engines
Agent Operational Directive
MANDATORY: Never output secret values or root/unseal material. Default to dynamic secrets and short TTLs. For Kubernetes auth, verify TokenReview permissions and reviewer JWT lifetime. On leak: revoke token/lease first, then rotate upstream credentials. Test policies with non-admin tokens before declaring least privilege done.
Source anchors (research)
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/vault/SKILL.md
Popularity
- Repository stars
- 1
- 7-day trend score
- 3
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/alivirgo/major-ai-skills/vault)