Infisical AI Skills

Give your AI coding agent accurate knowledge about Infisical — the open-source secret management platform.

Recommended: Connect our Docs MCP

The fastest way to stop your AI from hallucinating about Infisical is to connect our docs MCP server. It works with any MCP-compatible agent, auto-updates when our docs change, and requires zero maintenance.

URL: https://infisical.com/docs/mcp

Claude Code:

claude mcp add --transport http infisical-docs https://infisical.com/docs/mcp

Cursor / Windsurf: Add to your MCP settings:

{
  "mcpServers": {
    "infisical-docs": {
      "url": "https://infisical.com/docs/mcp"
    }
  }
}

VS Code / Copilot: Add to .vscode/mcp.json:

{
  "servers": {
    "infisical-docs": {
      "url": "https://infisical.com/docs/mcp"
    }
  }
}

Any MCP-compatible client can connect with that URL.

Alternative: Agent Skills

If your tool doesn't support MCP, or you want offline/local context, you can install these skills instead. They follow the Agent Skills open standard and work across 45+ AI tools.

Universal install

npx skills add Infisical/ai-skills

Claude Code (plugin marketplace)

/plugin marketplace add Infisical/ai-skills

Manual

Copy skill folders from skills/ into your project's agent skills directory:

Agent	Location
Claude Code	.claude/skills/
Codex	~/.codex/skills/
Cursor	.cursor/skills/ or .agents/skills/
GitHub Copilot	.github/skills/

What's included

17 skills across the whole Infisical platform. Each declares what it is not, so an agent lands on the right one — see AGENTS.md for the router and disambiguation table.

Secrets delivery

Skill	Covers
infisical-setup	CLI, all 9 SDKs (Node.js, Python, Go, Java, .NET, Ruby, PHP, Rust, C++), Docker, CI/CD, all 13 machine identity auth methods
infisical-api	/api/v4/secrets CRUD and batch, projects, identities, which endpoints paginate, rate limits
infisical-terraform	The provider's nested auth block, ephemeral resources, project roles, TFC OIDC
infisical-agent	Agent YAML config, Go template functions, sinks, polling, on-change commands
infisical-kubernetes-operator	v1beta1 CRDs and legacy v1alpha1, Helm install, auto-reload, push secrets

Moving and generating credentials

Skill	Covers
infisical-app-connections	All 83 connection types, auth methods, and credential fields (generated from source). The shared prerequisite for syncs, rotations, PKI, and scanning
infisical-secret-syncs	Pushing secrets to all 48 destinations, key schemas, initial-sync enums
infisical-dynamic-secrets	On-demand short-lived credentials across all 30 providers, leases, SSH certificates
infisical-secret-rotation	Rotating existing credentials across 28 providers, dual-phase vs single-phase, the two-user SQL pattern

Other products

Skill	Covers
infisical-pki	9 CA types, Policies/Profiles/Applications, API/ACME/EST/SCEP enrollment, 12 PKI Syncs, code signing, HSM, post-quantum
infisical-kms	Encrypt/decrypt, sign/verify, HMAC, key rotation, external KMS, KMIP, cosign
infisical-pam	Brokered access to 13 account types for humans and AI agents, session recording, JIT approvals
infisical-secret-scanning	GitHub/GitLab/Bitbucket data sources, infisical scan CLI, pre-commit hooks, honey tokens

Platform and governance

Skill	Covers
infisical-access-control	Roles, custom permissions, the granular secret actions, ABAC, temporary access, approvals, audit streams
infisical-sso	SAML/OIDC/LDAP and free Google/GitHub SSO, SCIM, group-to-role mapping, enforcement and break-glass
infisical-gateway	Private-network access with outbound-only tunnels, exact ports, Gateway Pools
infisical-self-host	Docker, Compose, Helm, env vars, Redis noeviction, FIPS 140-3, scaling and HA

Eval results

Every skill is A/B tested against a no-context baseline. We also ran a head-to-head comparison of Skills vs the Docs MCP. See evals/ for full data.

Skills vs no context

Skill	With Skill	Without	Delta
infisical-setup	100%	50%	+50pp
infisical-secret-syncs	100%	39%	+61pp
infisical-dynamic-secrets	94%	67%	+28pp
infisical-agent	100%	33%	+67pp

Skills vs MCP vs no context

Test case	No context	MCP (best-case)	Skills
Python SDK	0%	100%	100%
Node.js SDK	33%	100%	100%
API endpoints	38%	100%	100%
Terraform ephemeral	13%	100%	100%
Self-hosted Docker	38%	88%	100%
Average	24%	98%	100%

Both approaches dramatically reduce hallucination. The MCP is recommended because it auto-updates with the docs and requires no maintenance.

Accuracy audit: stale skills are worse than no skill

The skills are periodically re-verified against the Infisical codebase. The most recent audit ran a three-arm regression eval — no skill, pre-audit skill, post-audit skill — with tools disabled so the model could not look anything up:

Arm	Score	Pass rate
No skill	18/35	51.4%
Pre-audit skill	13/35	37.1%
Post-audit skill	35/35	100.0%

The pre-audit skills scored below the no-skill baseline. Outdated specifics don't merely fail to help — they override correct model knowledge. On the secret-syncs case the base model scored 5/5 unaided and the stale skill pulled it down to 2/5.

This is the strongest argument for the MCP: it tracks the docs automatically, so it cannot drift the way a vendored copy can. If you do install the skills, pin a version and re-pull when Infisical ships new providers or auth methods. Full data and a reproducible harness live in evals/accuracy-audit-2026-08/.

New skills

The 10 skills added in the 7 → 17 expansion, A/B tested the same way:

Arm	Score	Pass rate
No skill	27/56	48.2%
With new skill	56/56	100.0%

Null results are recorded, not hidden: on App Connections the base model already scored 4/4 unaided. The skills matter most where the model has little knowledge — PAM and SSO scored 1/5 unaided, and the Kubernetes Operator 0/5, because unaided it reaches for the legacy v1alpha1 InfisicalSecret CRD instead of current v1beta1. See evals/new-skills-2026-08/.

Generated reference files

The App Connection facts — 83 connections × auth methods × credential fields, and all 94 API endpoints — are generated from the Infisical source, not hand-maintained:

python3 tools/generate-app-connection-refs.py          # regenerate
python3 tools/generate-app-connection-refs.py --check  # CI: fail on drift

This is the structural answer to the drift problem above. Facts are derived; only guidance is written by hand. Re-verification is regenerate && git diff.

Routing: does the right skill get picked?

With 17 skills, mis-routing becomes the dominant failure mode — a skill loaded for the wrong question answers confidently from the wrong frame. So every skill declares what it is not, and AGENTS.md carries a router plus a disambiguation table for the pairs that get confused.

Measured on 14 prompts sitting deliberately on a seam between two similar skills:

Arm	Correct	Accuracy
Skill descriptions only	14/14	100.0%
With router + boundaries	14/14	100.0%

Getting there took two passes. Initially descriptions alone scored 13/14 — asked about short-lived SSH certificates it chose infisical-pam, plausible but wrong, since SSH certificates come from SSH dynamic secrets. The fix was moving the boundaries into the description frontmatter, because that is what actually decides whether a skill loads. See evals/routing-2026-08/.

Why this exists

AI coding agents frequently get Infisical details wrong:

What AI says	What's correct
pip install infisical-python	pip install infisicalsdk
from infisical_client import InfisicalClient	from infisical_sdk import InfisicalSDKClient
Use Service Tokens for Docker	Use machine identities (Service Tokens are deprecated)
npm install -g infisical	Install via apt from artifacts-cli.infisical.com
API Key Auth for Kubernetes	Kubernetes Auth (API Keys are deprecated)
GitHub syncs support importing	GitHub only supports overwrite (no import)
listSecrets(projectId, env, path)	listSecrets returns objects with .Key, .Value, .SecretPath fields
Agent uses JSON config	Agent uses YAML config with infisical: root key
require 'infisical-sdk' in Ruby	require "infisical" — gem name and require path differ
InfisicalSDK::InfisicalClient.new(url)	Infisical::Client.new(site_url: url)
Terraform provider "infisical" { client_id = ... }	Credentials nest inside auth = { universal = {...} }
Terraform ephemeral "infisical_secret" { secret_key = ... }	The attribute is name, not secret_key
GET /api/v4/secrets?offset=0&limit=20	That endpoint has no pagination; it returns everything at the path
Paginated responses return { items, total }	They return { <resource>, totalCount }
Self-hosted has no rate limits	Self-hosted has limits too (60 read / 200 write / 60 secrets per min by default)
GitHub sync scope environment	repository-environment; visibility is all/private/selected
import-prioritize-infisical	import-prioritize-source (values name source/destination, not the provider)
FIPS via infisical/infisical:latest-fips	FIPS 140-3 via the separate infisical/infisical-fips image
Kubernetes Operator uses InfisicalSecret	Current CRDs are v1beta1: InfisicalConnection, InfisicalAuth, InfisicalStaticSecret
A synced Kubernetes Secret restarts pods	It does not — add secrets.infisical.com/auto-reload: "true" to the workload
SQL rotation rotates one user's password	It alternates between two pre-existing users, username1 and username2
ECDSA P-256 in PKI is ECDSA_P256	The wire value is EC_prime256v1
GitHub secret scanning uses a github connection	It requires a github-radar connection
SSH certificates come from the PKI product	They come from SSH dynamic secrets
Reaching a private database requires self-hosting	It requires a Gateway; Infisical Cloud works fine
Any App Connection can use a Gateway	Only 16 of the 83 types accept gatewayId

These skills correct all of that.

Contributing

To add a new skill:

1. Create a directory under skills/ with a SKILL.md and optional references/ folder
2. Create a matching plugin wrapper under plugins/ with a .claude-plugin/plugin.json
3. Add a plugin entry in .claude-plugin/marketplace.json
4. Update AGENTS.md — add it to the router table, and add a row to the disambiguation table if it sits near an existing skill
5. Put the boundary in two places: a short "not for X (other-skill)" clause at the end of the description frontmatter — that is what decides whether the skill loads — and a ## Not this skill section in the body, with reciprocal rows on the neighbours it could be confused with
6. Run claude plugin validate . to check for errors
7. Add eval cases and run A/B benchmarks (see evals/ for examples)

Keeping skills accurate

Skill content is a vendored snapshot of a moving codebase, and the accuracy audit showed a drifted skill performs worse than no skill. When re-verifying:

• Cite the source, not the docs prose. Counts and enum values come from the code: secret-sync-enums.ts, dynamic-secret/providers/models.ts, db/schemas/models.ts (IdentityAuthMethod), server/routes/v4/. Docs pages lag; enums don't.
• Prefer exact literals over prose descriptions. repository-environment beats "the environment scope". Wrong literals are the failure mode that hurts most.
• Re-sync the plugin wrappers. plugins/<name>/skills/<name>/ is a copy of skills/<name>/; they drift silently. Diff them before committing.
• Run the regression eval. evals/accuracy-audit-2026-08/run_evals.py compares no-skill / old-skill / new-skill with tools disabled. Add assertions for whatever you just corrected so the next audit catches a regression.

License

MIT