Launch Week 3: Five days of launches

Skill

Overview

The Confident AI SDK exposes every Skill method on the platform. This page documents how to call these methods in all supported languages. See the introduction to install the SDK and set your API key.

Methods

Get Governance Policy Skill

Retrieves the Agent Skill attached to a governance policy — not a control, and nothing about it is assessed: it is the instructions Confident AI serves to coding agents working on the projects this policy governs. A project enrolled in this policy finds it in its Agent Skills git repository as skills/governance/SKILL.md. Answers with null when the policy has none, in which case agents receive no governance skill.

from confident_ai import ConfidentAI

client = ConfidentAI()

result = client.organization.get_governance_policy_skill(
    policy_id="<GOVERNANCE-POLICY-ID>",
)

For async mode, call a_get_governance_policy_skill and await it as shown below:

result = await client.organization.a_get_governance_policy_skill(...)

Parameters

ParameterTypeDescription
policy_idstrRequired. The id of the governance policy.

Returns

This method returns an object of type GovernancePolicySkill.

Update Governance Policy Skill

Sets the Agent Skill served to coding agents working on the projects this governance policy governs, creating it when the policy has none and replacing it outright otherwise. Both description and body are required, so there is no way to patch one and leave the other. The new text reaches an agent the next time it clones the project's Agent Skills repository.

from confident_ai import ConfidentAI

client = ConfidentAI()

result = client.organization.update_governance_policy_skill(
    policy_id="<GOVERNANCE-POLICY-ID>",
    description="Rules for shipping changes to an AI system governed for EU AI Act readiness.",
    body="## Before opening a pull request\n\n- Run the project's evals and attach the test run link.\n- Never disable a governance control to make a build pass.\n",
)

For async mode, call a_update_governance_policy_skill and await it as shown below:

result = await client.organization.a_update_governance_policy_skill(...)

Parameters

ParameterTypeDescription
policy_idstrRequired. The id of the governance policy.
descriptionstrRequired. One line on what the skill covers, which is how a coding agent decides whether it is relevant. Required, and must not be blank.
bodystrRequired. The instructions themselves, in Markdown, which become the body of skills/governance/SKILL.md. Required, and must not be blank.

Returns

This method returns an object of type GovernancePolicySkill.

Delete Governance Policy Skill

Permanently deletes the Agent Skill attached to a governance policy. Coding agents working on the projects it governs stop receiving skills/governance/SKILL.md on their next clone. The policy, its controls and its projects are untouched, since the skill instructs agents rather than gating anything. A policy that has no skill is reported as not found.

from confident_ai import ConfidentAI

client = ConfidentAI()

result = client.organization.delete_governance_policy_skill(
    policy_id="<GOVERNANCE-POLICY-ID>",
)

For async mode, call a_delete_governance_policy_skill and await it as shown below:

result = await client.organization.a_delete_governance_policy_skill(...)

Parameters

ParameterTypeDescription
policy_idstrRequired. The id of the governance policy.

Returns

This method returns an object of type GovernancePolicyIdentifier.

Types

GovernancePolicyIdentifier

A governance policy, identified by its id.

class GovernancePolicyIdentifier:
    id: str

idstrRequired

The id of the governance policy.

Example: "<GOVERNANCE-POLICY-ID>"

GovernancePolicySkill

An Agent Skill attached to a governance policy: the instructions Confident AI serves to coding agents working on the projects the policy governs. Confident AI publishes a per-project Agent Skills git repository, and a project enrolled in this policy finds the skill there as skills/governance/SKILL.md, with description in the YAML frontmatter and body as the Markdown beneath it. Nothing about the skill is assessed — it instructs the agent, it is not a control.

class GovernancePolicySkill:
    description: str
    body: str

descriptionstrRequired

One line on what the skill covers, which is how a coding agent decides whether it is relevant to the task in front of it.

Example: "Rules for shipping changes to an AI system governed for EU AI Act readiness."

bodystrRequired

The instructions themselves, in Markdown. This becomes the body of the SKILL.md file, so write it for a coding agent rather than a person and be as explicit as the policy requires.

Example: "## Before opening a pull request\n\n- Run the project's evals and attach the test run link.\n- Never disable a governance control to make a build pass.\n"

Building a production pipeline?Design a scalable API workflow for evals, datasets, traces, and promptsTalk to an engineer

Last updated on

Built byConfident AI