prime-agent-feature-factory

Prime Agent integration for feature-factory. It adds a /feature command and a Prime-specific adapter skill while preserving the factory package's canonical workflow and CLI-owned durable state.

Install

prime-agent package install npm:prime-agent-feature-factory

The package manifest exposes its extension and skill through pi.extensions and pi.skills. Prime Agent installs the runtime feature-factory dependency with it. Node.js 22 or newer is required.

Use

/feature [--autonomous | --headless] <ticket key | feature idea>

Unattended, a run needs two things beyond the invocation, both documented with their reasoning in the repository's operator guide: redirect stdin from /dev/null, because an inherited pipe that never reaches EOF blocks the CLI before it emits a byte and looks exactly like a broken tool; and raise the --autonomous limits, because the defaults of 12 turns, 3 continuations, 80,000 tokens and 30 minutes stop a real run mid-flight in a way that looks like a stall. -p alone prints one response and exits, which initializes a run and then abandons it.

The integration currently drives foreground runs only. It rejects --background before creating or changing a run. The extension exposes the current Prime session and installed agent directory through feature_factory_context; the skill uses Prime RLM children for specialist work and requires their reports through agent messaging.

skills/feature/WORKFLOW.md is a package-local copy of the factory's canonical workflow. pretest and prepack refresh it with the repository sync script so the adapter never relies on a skill loader to inline another package's resource. The CLI remains the only writer of run.json.

Specialist model and thinking level

Every agent file in feature-factory declares model, role, and effort in its frontmatter, and the OpenCode adapter treats the three differently: role selects a configured profile, effort becomes the default variant, and the declared model is ignored — an OpenCode agent's model comes from profile configuration, because sonnet is a tier rather than a model id. This adapter consumes none of the three. On Prime, all eleven specialists run with the parent session's model and thinking level.

That follows from Prime's spawn contract rather than being an oversight. rlm.run accepts exactly two options — name and model — and unknown keyword arguments fail the spawn instead of being ignored, so an effort or thinking argument would stop the child starting rather than go unused. A child inherits the parent model when model is omitted, and inherits the global defaultThinkingLevel either way.

model is deliberately not passed even though Prime supports it. Selection is fail-closed: an unavailable model fails admission rather than falling back to another one. Under a subscription OAuth login (PrimeIntellect-ai/prime-agent#740) rlm.find_models() returns nothing and an explicit model= fails admission for every model except the parent's, so passing a selector there would stop specialists spawning at all — worse than running them all on one tier. Pre-validating a selector does not avoid this either: find_models() returns an alphabetical head slice that reads as the full reachable set (#799), so it rejects models that are in fact reachable.

The parent session is therefore the only lever. work-decomposer and work-reviewer declare opus and decide whether a plan is satisfiable and whether a build is accepted; on Prime they inherit whatever the run was launched with. Choose the parent model and defaultThinkingLevel with that in mind.

Upstream requests that would change this: per-child thinking level (#703), a persistent subagents.defaultModel policy (#921), and virtual model selectors (#1138). If they land, what this adapter needs is mostly configuration rather than code.

Development

From this package directory:

npm test
npm pack --dry-run

License

MIT