companion-inc

Predict or audit protein structures with AlphaFold2-style workflows. Use when a research task needs monomer/multimer structure prediction, MSA/template handling, confidence metrics, or comparison against PDB/AlphaFold references.

Run or plan Boltz biomolecular structure predictions for proteins, complexes, ligands, or nucleic-acid assemblies. Use when a task asks for Boltz setup, inputs, outputs, confidence interpretation, or reproduction.

Use Borzoi-style regulatory genomics models for sequence-to-expression or variant-effect analysis. Use when the task asks for noncoding variant impact, regulatory sequence design, or expression prediction.

Run or prepare Chai-1 structure predictions for biomolecular complexes. Use when a task asks for Chai-1 inputs, multimers, ligand/nucleic acid structure prediction, or confidence review.

Set up a reproducible Feynman compute environment for research jobs. Use when a task needs Python/R packages, GPU libraries, containers, Modal, SSH, caches, or managed model runtime setup.

Configure Feynman specialists, skills, connectors, permissions, memory categories, compute providers, and project setup. Use when the task asks to customize the research workbench or create a reusable Feynman research capability.

Run or plan DiffDock molecular docking workflows. Use when a task asks for protein-ligand pose prediction, docking setup, ligand/protein preparation, pose ranking, or docking-result verification.

Predict quick protein structures with ESMFold-style workflows. Use when a task needs fast MSA-free folding, sequence triage, variant structure screening, or confidence review.

Set per-plan caps, quotas, limits, and compute sizes in Companion's billing tiers by deriving the numbers from the existing price/credit ladder, not by inventing round or evenly-doubling numbers. Load when adding or changing a per-tier quantity (automations cap, storage, rate limit, feature quota, or per-plan compute/runtime sizing like vCPU/RAM or sandbox template) or editing PLAN_TIERS / COMPUTER_RUNTIME_TIERS in src/lib/billing/constants.ts. Not for one-off non-tiered constants.

Run Introspect for Codex sessions when the user asks Introspect to inspect recent corrections, update agent instructions, or report what it would remember.

How an LLM agent should emit a structured decision or host-consumed payload — a mode flag, control action, state change, routing choice, or UI data envelope — back to its host system. Use when building or debugging an agentic product where the host asks the model to return only JSON, signal quiet/silent/archive/done/hand-off state, choose an app action, feed a menu, or when markers/tags leak into the transcript or host code regex-scrapes prose.

Drive authenticated web-console and browser forms through semantic controls, Playwright/DOM locators, or form APIs instead of repeated coordinate clicks and blind keystrokes. Use when filling or editing forms in Chrome or the in-app browser, especially admin consoles, OAuth/client settings, secret managers, or pages where focus or typing is failing. Not for read-only page inspection or a single obvious click.

Retrieve an API token, secret, GitHub credential, or remote-session auth path from the machine's own stores before asking the user to paste it or working around auth with file transfer. Use when a CLI errors on auth, an env var like SENTRY_AUTH_TOKEN is unset, a deploy/API call needs a key, or a remote SSH shell cannot access credentials that exist in the GUI/session. Not for a secret that exists nowhere and must be freshly minted through a human-only step (hardware MFA, a one-time email/SMS code) — drive that flow per AGENTS.md instead.

Register a folder as an Obsidian vault before handing off `obsidian://open` links on macOS, so the user actually opens what you generated instead of hitting "Unable to find a vault for the URL".

Polish user-facing product surfaces such as pricing pages, plan cards, feature pages, grant/application packets, navigation rows, empty states, model/settings pickers, provider import/export prompts, provider logos, App Store/TestFlight metadata, icons, categories, onboarding, and marketing UI by reading real reference products and the product's enforced source of truth before writing copy, visuals, budgets, or interaction structure. For research grant packets, keep funder-required research intent distinct from product/prototype framing.

Create, update, prune, and validate agent-loop skills as scoped procedural memory instead of bloating AGENTS.md. Use for reflector decisions involving skill_new, skill_update, skill_prune, reusable workflows from observed failures, or placement decisions between a skill and the global prompt.

Reproduce a bug with a failing test before fixing it, so the fix is proven and can't silently regress. Use when the user reports a bug, says "this is broken", "X returns the wrong value", "fix this crash", or hands over a stack trace or a failing case. Write the smallest test that fails for the stated reason, then fix until it passes — and only it changes from red to green.

Turn a messy working tree into clean, reviewable commits and a tight PR description. Use when the user says "open a PR", "commit this", "clean up my commits", "get this ready to push/review", or when a dirty tree mixes unrelated changes. Splits work into one logical change per commit, writes conventional messages, and drafts a PR body grounded in the actual diff.

Orient yourself in an unfamiliar repository before doing any work. Use when you open a new project, clone a repo, switch into a codebase you don't know, or ask "how do I run this / where do I start / how is this project set up". Produces a concrete first-run checklist (install, build, test, run) and a short map of the codebase, grounded in the project's own files — never guessed.

Capture a compact, resume-ready checkpoint of the current Codex session so work survives a long thread, a context reset, or a handoff. Use when the conversation is getting long, before switching tasks, when the user says "checkpoint", "save where we are", "summarize so we can continue later", or "hand this off". Writes a durable status note — not chat fluff — that the next session can act on cold.

Create BetterCodex plugins and themes for marketplace submission, or Codex skills for a Codex-native plugin bundle.

Install local BetterCodex plugins and themes into the desktop runtime folders.

Offload a Claude Code task to a remote Companion sandbox via SSH. Use when the user wants to continue a task in the cloud, run something remotely, or offload work to their Companion sandbox. Triggers on "offload", "run in cloud", "companion run", "remote task", or "send to sandbox".