Source: docs/integration/use-with-an-agent.md
Use Forgium Agent with an agent
This page is the universal entrypoint for a coding agent working in your host
application. The prompt below is intentionally portable: it works when the
agent can read a public documentation URL and does not require an Agent Skills
runtime.
Choose a workflow
- Universal mode — recommended: copy the prompt below and give it to your
coding agent.
- Guided mode: use the public
Forgium Agent Integrator Skill when
your agent runtime supports Agent Skills. The Skill is optional and is a
workflow that points back to this documentation; it is not a second API
contract.
There is currently no universal installation command or package registry for
Skills. Do not run a guessed command. Use the prompt unless your runtime has a
supported, verified way to load a SKILL.md file.
Copy this prompt
You are integrating my application with Forgium Agent. Use only the official
public integration documentation at https://docs.forgium.dev/docs.
Read these pages in order before changing code:
1. https://docs.forgium.dev/docs/integration/overview.html
2. https://docs.forgium.dev/docs/integration/getting-started.html
3. https://docs.forgium.dev/docs/integration/environments.html
4. Choose the relevant HTTP, capabilities, and API reference pages.
5. When the runtime supports Agent Skills, read the Forgium Agent Integrator Skill at
https://docs.forgium.dev/docs/skill/micro-agent-integrator/SKILL.html. Treat it
as the execution workflow; the public documentation remains the canonical contract.
Before changing code, confirm that these values are available for the target
environment:
- FORGIUM_AGENT_BASE_URL
- FORGIUM_AGENT_API_KEY
If either value is missing, ask the Forgium administrator to provide it. Use
only the public API and documentation. Do not request platform-operator
credentials or access to internal services.
First perform the authenticated HTTP smoke test. Inspect this repository and
choose the integration path it actually needs:
- HTTP for a synchronous host-application response; or
- capabilities when the agent needs approved live business data or actions.
If capabilities are needed, follow the complete public capability lifecycle:
create a scoped HTTPS host endpoint, provision its credential through the
Public API vault, create a secret-free manifest, import it, contract-test it,
activate it, and verify the result through chat. Repeat this separately for
each environment. Never use a placeholder URL or an unauthenticated endpoint.
Implement the integration in this repository, run its tests, and verify an
end-to-end request. Report exactly one concrete blocker if an authorized
endpoint deployment or HTTPS exposure step is unavailable. Do not claim that a
capability is active unless activation and the final chat verification pass.
For staging validation, replace https://docs.forgium.dev/docs with
https://docs-staging.forgium.dev/docs. The administrator provides the API key
out of band; do not put it in this document, a prompt committed to source
control, a manifest, or logs.
Expected result
A successful agent run leaves a tested first message and a documented choice of
HTTP or capabilities. For capabilities, completion additionally means that the
target-environment capability is active and the final chat response is grounded
in the approved host endpoint.
Continue with: