rat-stack-mode
Skill2 starsChanged 36 days agoMITView on GitHub
Route non-trivial rat-stack code, data, design, investigation, and documentation tasks to every matching playbook and principle.
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- add-a-capabilitySkill
- add-a-lifecycle-machineSkill
- add-a-storeSkill
- gardenerSkill
- investigate-a-failureSkill
- keep-or-cutSkill
- learn-alchemySkill
- profile-a-workerSkill
- query-designSkill
- shipSkill
- test-behaviourSkill
- uncomplectSkill
- write-a-wiki-pageSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
Start a todo list with reading the principles index below.
- Distinctive: line 108
Follow AGENTS.md's Boundaries and sign-off for autonomy.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Work in rat-stack mode
AGENTS.md is law. VISION.md sets the intent, reliability goals, scope, package-removal test, and shrinking debt. Use this skill to choose playbooks under those rules.
A capability is one named action with shared value definitions and a server-side implementation. A cartridge is a package for one job with its implementation and required infrastructure.
Non-negotiables
- Start a todo list with reading the principles index below.
- Open every leaf that bears on a decision. Report which rules changed which choices.
- Name the data shape, authoritative store, copies calculated from it, and writer before writing code.
- Reuse stores when their facts and operation guarantees fit. Reporting does not justify gating unrelated execution.
- Measure before setting a timeout, rate cap, batch size, or concurrency limit. Put the baseline first; requested values remain hypotheses.
Principles index
Leaves live in the repo; read them from a checkout.
Product and UX
skills/rat-stack-mode/principles/judge-through-the-person.md: changing an interface or user-visible behavior.skills/rat-stack-mode/principles/keep-boundary-words-honest.md: naming failures, outcomes, and public claims.
Architecture
skills/rat-stack-mode/principles/one-contract-many-projections.md: adding actions for agents or browser clients.skills/rat-stack-mode/principles/keep-the-stack-floor.md: choosing runtime patterns, infrastructure, and validation.skills/rat-stack-mode/principles/prove-move-delete.md: replacing an internal API or implementation.skills/rat-stack-mode/principles/use-real-provider-guarantees.md: choosing an adapter or testing its behavior.skills/rat-stack-mode/principles/gates-guard-invariants.md: adding a gate or approval check.skills/rat-stack-mode/principles/keep-unknown-distinct.md: handling crashes, refusal, or missing results.
Data and truth
skills/rat-stack-mode/principles/name-the-system-of-record.md: adding stores, caches, aggregates, or nightly reports.skills/rat-stack-mode/principles/store-source-facts.md: capturing observations or interpreting missing fields.skills/rat-stack-mode/principles/keep-execution-receipts.md: recording runs or claiming an operation succeeded.skills/rat-stack-mode/principles/scope-private-data.md: handling identities, credentials, logs, or analytics.skills/rat-stack-mode/principles/version-content-artifacts.md: writing public content or quoting code.skills/rat-stack-mode/principles/no-dual-writes.md: copying a fact across stores or maintaining a read model.skills/rat-stack-mode/principles/validate-before-durable-write.md: recording rows, events, or snapshots.skills/rat-stack-mode/principles/bound-every-read.md: listing, querying, exporting, or rebuilding data.skills/rat-stack-mode/principles/one-writer-per-partition.md: updating shared mutable state.skills/rat-stack-mode/principles/make-operations-idempotent.md: replaying commands or recovering after crashes.skills/rat-stack-mode/principles/subtract-before-you-add.md: sequencing new work or removing unused interfaces.
Verification
skills/rat-stack-mode/principles/check-the-actual-artifact.md: finishing work or reviewing another agent's result.skills/rat-stack-mode/principles/prove-the-recipient-path.md: changing runtime wiring or deployed interfaces.skills/rat-stack-mode/principles/tests-detect-broken-behavior.md: adding tests or judging their coverage.skills/rat-stack-mode/principles/find-the-cause-first.md: fixing failures or slow runs.skills/rat-stack-mode/principles/rebuild-what-you-run.md: running or debugging an artifact someone else built.skills/rat-stack-mode/principles/measure-real-limits.md: setting budgets or diagnosing contention.
Agents
skills/rat-stack-mode/principles/look-it-up-first.md: entering unfamiliar code or deciding whether to ask.skills/rat-stack-mode/principles/ask-one-material-question.md: reaching a decision that needs owner sign-off.skills/rat-stack-mode/principles/separate-writers-verify-readers.md: dividing work or accepting delegated results.skills/rat-stack-mode/principles/encode-lessons-in-structure.md: correcting a repeated bad pattern.
Prose and visual
skills/rat-stack-mode/principles/use-plain-language.md: writing docs, skills, or system pages.skills/rat-stack-mode/principles/design-serves-function.md: changing layout, focus, or visual emphasis.skills/rat-stack-mode/principles/make-review-views-complete.md: rendering evidence for people and agents.
Cost and speed
skills/rat-stack-mode/principles/optimise-the-whole-loop.md: changing builds, tests, generation, or feedback time.
Playbooks
Match every applicable playbook; a capability with new persistence matches capability and store playbooks. Open each match and copy its steps into the todo list before task-specific steps. For unnumbered skills, copy action lists; keep skipped steps as skip: <reason>. Finish with each matching playbook's report requirements.
- New action: add-a-capability.
- States, retries, or cancellation: add-a-lifecycle-machine.
- New persistence: add-a-store.
- Public prose or skill pages: write-a-wiki-page.
- Package or interface choices: keep-or-cut.
- Stateful design, replacement, or read-model design: uncomplect.
- Pins, recurring bad patterns, or repository maintenance: gardener.
- Cloud resources or unfamiliar deployment wiring: learn-alchemy.
- Unfamiliar stack integration: read the rat-stack template at https://ratstack.sh/llms.txt.
- Crashes, flaky tests, wrong output, or slow runs: investigate-a-failure.
write-a-query, add-a-read-model, and explore-limits are planned. Use applicable leaves and existing playbooks; identify uncovered steps explicitly.
Autonomy and child-project slots
Follow AGENTS.md's Boundaries and sign-off for autonomy. In a child project, name production write authority, deploy gates, and the model roster here and in AGENTS.md.
Router and adapted principles credit pstack by Lauren Tan (MIT, skills/rat-stack-mode/LICENSE-pstack).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

