sno-station-skills

Prove that a reported run caused its intended real effect. Walk the executor path from entry to effect, capture one observable per hop, and report a result a no-op could not produce.

Write, revise or review a charter: the one document that tells an executor what to deliver, what is out of bounds, what the owner already decided, and how each success check will be proven. Use when work needs a written brief before someone else does it, or when asked to write, change or check a charter.

Check external services, hosts, browser sessions, model endpoints, and agents before an end-to-end run. MANUAL-ONLY: load when a caller names it. NEVER auto-load from context for ordinary testing or debugging.

Classify one unexpected end-to-end test failure as a product defect or an environment fault needing a new preflight check. MANUAL-ONLY: load when the suite runner names a specific failure. NEVER auto-load from context.

MANUAL-ONLY: challenge a product idea, design, PRD, or architecture from first principles. Use only when the user names this skill. Research first, then write a local review and PRD handoff.

Write the least code that works, and cut what an AI agent over-built. Use on any coding task: write, fix, refactor, or review a diff. Also when the user says 'less', 'simplify', 'too complex', 'over-engineered', or 'what can we delete'. Not for prose, research, or non-code questions.

Check agent-produced plans, tests, or code against the owner's actual intent. Find missing meaning, unrequested work, stale truth, unreachable decisions, unsupported completion claims, and implementation debt. Use before approval or merge.

Implement and refactor production Python, FastAPI, Pydantic, async, and migrations.

Implement, refactor, and migrate production Rust with repository-grounded ownership, type-driven invariants, explicit errors, safe concurrency, minimal unsafe code, and verified Cargo workflows. Use for Rust features, crates, CLIs, services, libraries, async code, FFI, performance work, migrations, or substantial .rs and Cargo.toml changes, including requests such as 'rust-coder', 'write Rust', or 'implement this Rust feature with Codex'.

Implement and refactor production shell scripts, command-line tools, automation, process wrappers, and Git or infrastructure helpers with reliable failures, useful logs, and machine-readable output.

Write and improve test-owned code: tests, test scripts, fixtures, evals, and test-only helpers. Use whenever Codex is asked to create or change test code. Its only job is to produce the strongest practical tests in the repository's existing harness, run them, and fix defects in the test code.

Implement and refactor production TypeScript/TSX, strict types, APIs, and migrations.

Answer how long agent work will take, how much longer active work needs, or what clock time applies. Estimate full AI-agent wall time, measure active progress first, and get owner-facing times from `sno report-time`. Never estimate human-developer time.

Give the owner one page when they come back from sleep, a meeting or a long run: what got done, what is stuck, what waits for their answer, and what it cost in quota. Use when the owner asks what happened while they were away, wants a status page, or when you finish a long stretch of work and must report it.

Report how many problems the two brains found: mutual review (the other model checks the work, from the peer-review records) and self-check (an agent finds its own error in its own conversation, judged by that agent's own model), side by side. Use when the owner asks whether the second agent or the reviews are worth it, or wants the week's review results.

Adversarial review and fix loop for code, specs, plans, and migration manifests.

Check that this machine's agent team can work and say what is wrong in plain lines: tools installed, core skills installed, skill commands runnable, hooks configured, Reach seat live, a heartbeat running, both vendors' quota readable, temp space free. Checks only, never repairs. Use on a new machine, when agents stop reaching each other or a long run fails to start, or when the owner asks whether everything is healthy.

Operate the `sno` command-line tool: install, update, check, and uninstall Sno products, read the exact commands and flags of the installed version, and run the account, skills, onboarding, Station telemetry and REM commands. Use when the owner or a task says "sno", "Sno CLI", "sno setup", "sno update", "sno doctor", "sno uninstall", "sno station", "telemetry consent", "rem verdict", "sno account", or "sno skills". The exact flags always come from the installed binary, never from this file. Do not use it to decide the owner's choices for them.

Self Learning from this machine's Claude Code, Codex, Hermes, OpenClaw and Cursor sessions. Decide skill changes and lessons with accept/reject/tbd; recall lessons at session start.

Transfer full task ownership to a fresh agent with a verified brief and explicit release. Use for replacing the current executor, including a directly assigned task with no work card; not for supervising a helper.

Set recurring checks or completion watches with `sno heartbeat` and its persistent log reader. Use for periodic reports or unattended long runs. Not for calendar scheduling or commands watched directly.

Register this agent and terminal with Reach for cards, calls and wake-ups. Use when asked to join, connect, or become reachable, or when a manually opened agent lacks `SNO_EXECUTOR_ADDR`.

Contact another agent, exchange work cards, inspect replies, or manage reachable seats through the installed Reach program. Use for communication; full task ownership transfer belongs to handoff.

Quota-triggered failover from Claude Code, Codex, Cursor, Hermes or OpenClaw. When armed, it reads the sender quota and tries other-vendor models in place, then receiving agents in the configured order. Use when a long run must finish even if one vendor's quota runs out. Not a monitor for anything else.

Read remaining Claude Code and Codex subscription quota without spending it. Use when the user asks or after a vendor rejects work for quota; a caller such as rotate-agent may also arm a scheduled read on purpose. Never poll on your own; after a `wait` verdict, pass the reset wait to `sno heartbeat` and re-read once when it ends.

Carry out a released charter to a proven result: plan only as much as the work needs, do the work through the project's own workers, prove every success check with a recorded real run, get one independent review, and close honestly. Invoke explicitly as `$deliver <charter>` (Codex) or `/deliver <charter>` (Claude, Cursor), or the way your agent invokes a skill, or when a dispatch names you the executor of a charter.

Route production coding tasks from Claude Code to the user-level Codex Shell, TypeScript, Python, or Rust workflow through a direct `codex exec` run. Use for "codex-coder", "ask Codex to write code", "generate with Codex", language migrations, or mixed Shell/TypeScript/Python/Rust implementation.

Route production coding tasks to the user-level Shell, TypeScript, Python, or Rust Codex workflow. Use for "codex-coder", "Codex write", "generate with Codex", language migrations, or mixed Shell/TypeScript/Python/Rust implementation when Codex should choose the correct language specialist automatically.