@ffaerber — before I draft a plan, I need a few specifics since "intern" isn't just a config toggle — it needs a role definition like the others, and several things in agent.yml intentionally couple to a fixed agent list:
Role/responsibilities: What should @intern actually do? @junior already runs on the local ornith model for "small, low-risk YAML/compose/config" changes — if @intern uses the same model, what's the differentiation (e.g. docs-only, tests-only, "even smaller/no-PR" scope, or just extra capacity/parallel junior)?
Mode: Should @intern open PRs like @junior/@senior/@lead (mode: pr), or comment-only like @pm/@qa (mode: comment)?
Vision: Ornith is text-only (same as @junior), so I'll assume vision: false unless you say otherwise.
Own Gitea identity: Do you want a TOKEN_INTERN secret/bot user (so commits/PRs are authored as intern), or is falling back to the shared bot token fine for now?
Why this matters technically: adding a new name isn't a one-line edit. It requires touching several tightly-coupled spots in agent.yml that AGENTS.md explicitly calls out as sensitive (the trigger gate, delegation-target regex used in multiple places, and the token-selection case blocks) — plus the README's roster/secrets table. Getting the role/mode wrong means redoing all of those, so I'd rather confirm intent first.
If @intern is meant to just be "a second @junior-style low-risk YAML/config dev," say so and I'll skip straight to planning with that assumption.
🤖 **@pm**
@ffaerber — before I draft a plan, I need a few specifics since "intern" isn't just a config toggle — it needs a role definition like the others, and several things in `agent.yml` intentionally couple to a fixed agent list:
1. **Role/responsibilities**: What should `@intern` actually do? `@junior` already runs on the local `ornith` model for "small, low-risk YAML/compose/config" changes — if `@intern` uses the *same* model, what's the differentiation (e.g. docs-only, tests-only, "even smaller/no-PR" scope, or just extra capacity/parallel junior)?
2. **Mode**: Should `@intern` open PRs like `@junior`/`@senior`/`@lead` (`mode: pr`), or comment-only like `@pm`/`@qa` (`mode: comment`)?
3. **Vision**: Ornith is text-only (same as `@junior`), so I'll assume `vision: false` unless you say otherwise.
4. **Own Gitea identity**: Do you want a `TOKEN_INTERN` secret/bot user (so commits/PRs are authored as `intern`), or is falling back to the shared bot token fine for now?
Why this matters technically: adding a new name isn't a one-line edit. It requires touching several tightly-coupled spots in `agent.yml` that AGENTS.md explicitly calls out as sensitive (the trigger gate, delegation-target regex used in multiple places, and the token-selection `case` blocks) — plus the README's roster/secrets table. Getting the role/mode wrong means redoing all of those, so I'd rather confirm intent first.
If `@intern` is meant to just be "a second `@junior`-style low-risk YAML/config dev," say so and I'll skip straight to planning with that assumption.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
please create me a new agent called intern he is using the local ollama model ornith
@pm here
🤖 @pm
@ffaerber — before I draft a plan, I need a few specifics since "intern" isn't just a config toggle — it needs a role definition like the others, and several things in
agent.ymlintentionally couple to a fixed agent list:@internactually do?@junioralready runs on the localornithmodel for "small, low-risk YAML/compose/config" changes — if@internuses the same model, what's the differentiation (e.g. docs-only, tests-only, "even smaller/no-PR" scope, or just extra capacity/parallel junior)?@internopen PRs like@junior/@senior/@lead(mode: pr), or comment-only like@pm/@qa(mode: comment)?@junior), so I'll assumevision: falseunless you say otherwise.TOKEN_INTERNsecret/bot user (so commits/PRs are authored asintern), or is falling back to the shared bot token fine for now?Why this matters technically: adding a new name isn't a one-line edit. It requires touching several tightly-coupled spots in
agent.ymlthat AGENTS.md explicitly calls out as sensitive (the trigger gate, delegation-target regex used in multiple places, and the token-selectioncaseblocks) — plus the README's roster/secrets table. Getting the role/mode wrong means redoing all of those, so I'd rather confirm intent first.If
@internis meant to just be "a second@junior-style low-risk YAML/config dev," say so and I'll skip straight to planning with that assumption.