mirror of https://github.com/hashicorp/agent-skills.git synced 2026-09-17 14:17:29 +08:00

Files

T

Lee Fowler 4451ceca54 feat: consolidate governance implementation and skill updates (#90 )

* docs: improve token-efficient terraform state access

* feat: add provider configuration skill

* refactor: align provider resources with plugin framework

* refactor: generalize provider actions guidance

* feat: add provider scaffold and authentication guidance

* docs: add acceptance test environment guidance

* docs: register provider documentation and test skills

* docs: expand provider documentation guidance

* feat: add provider framework migration skill

* feat: add provider ephemeral resources skill

* feat: consolidate product plugins and apply governance

* fix: correct imported skill metadata and references

* docs: update supported gpt model baselines

* fix: preserve unreleased changelog entries

* ci: remove tessl skill evaluations

* ci: streamline repository validation

* docs: streamline contribution guidance

* docs: remove migration prose from changelog

* docs: refine issue intake and security guidance

* docs: refine public repository guidance

* docs: add readme navigation and recent updates

* docs: update wording in skill catolog file

---------

Co-authored-by: Andy Baran <andy.baran@hashicorp.com>
Co-authored-by: Tyler Lynch <tyler.lynch@hashicorp.com>

2026-08-10 11:37:46 -04:00

2.1 KiB

Raw Permalink Blame History

Agent Instructions

This repository contains public HashiCorp Agent Skills and the Terraform and Packer product plugins that distribute them.

Canonical structure

Put Terraform Skills in plugins/terraform/skills/<skill-name>/.
Put Packer Skills in plugins/packer/skills/<skill-name>/.
Keep Claude Code and Codex plugin manifests at each product root.
Keep the Claude Code and Codex marketplaces aligned to exactly the terraform and packer bundles.
Keep individual Skill installation paths in SKILLS.md aligned with disk.

Skill requirements

Every Skill directory must contain SKILL.md.
Every SKILL.md frontmatter must contain name, description, and metadata.lifecycle-status.
Skill names must match their directory names.
Supported lifecycle states are active, deprecation-candidate, deprecated, and retired.
Keep detailed material in directly linked references/, scripts/, or assets/ resources when that improves progressive disclosure.

Ownership and review

CODEOWNERS is canonical. Every Skill must have an explicit Skill-level entry that includes @hashicorp/team-agent-skills-ecosystem. Product-aligned owners may be added only after their participation is confirmed. The ecosystem team owns repository conventions, evaluation integration, distribution, supported model consistency, and final repository decisions.

Review or reevaluate a Skill whenever the Skill changes or the supported model matrix changes. Structural validation, Skill review, private Waza evaluation, CODEOWNERS review, and maintainer judgment are complementary inputs.

Contributions

The repository is in an internal-contribution-only phase. Follow CONTRIBUTING.md and use the proposal template for a new Skill or substantial rewrite. Do not add public evaluation fixtures or raw evaluation results.

Validation

Run these checks before proposing a change:

./scripts/validate-structure.sh
git diff --check

Run relevant example, link, installation, and harness checks for the changed surface. Never run infrastructure-creating examples against real credentials as part of repository validation.

2.1 KiB Raw Permalink Blame History