distill-to-rutters
Skill4 starsChanged 4 months agoMITView on GitHub
Use when an existing Markdown skill instruction should be transformed into transparent Rutters and an operable Voyage dispenser. Do not use for ordinary summarization, general Rutter development, or operating an existing compass.
In this repository45 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- bib-auditSkill
- bootstrap-dispatcher-runtimeSkill
- ci-debugSkill
- cloud-filesSkill
- connect-googleSkill
- daily-planSkill
- email-clientSkill
- email-triageSkill
- find-handoff-candidatesSkill
- formal-prose-reviewSkill
- get-weatherSkill
- git-workflowSkill
- hook-makerSkill
- initialize-tddSkill
- install-launchersSkill
- latex-workshopSkill
- list-managerSkill
- llm-wakeupSkill
- loose-modeSkill
- make-tex-docstringSkill
- math-dependency-graphSkill
- milestone-loggingSkill
- node-certifySkill
- node-driftSkill
- notation-reviewSkill
- online-calendarSkill
- pdf-to-markdownSkill
- prepare-handoffSkill
- proof-auditSkill
- recurring-tasksSkill
- refactor-nodeSkill
- regenerate-blueprintsSkill
- relocate-nodesSkill
- semantic-integrationSkill
- send-feedbackSkill
- setup-interface-managerSkill
- skill-makerSkill
- technical-flow-reviewSkill
- tight-modeSkill
- tool-applicabilitySkill
- update-standardsSkill
- using-compassSkill
- wrap-upSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 58
Every stage artifact must live in that exact <source-stem>_distillation workspace and preserve one common root source prerequisite.
- Distinctive: line 75
Before every post-bootstrap route, invoke that injected interface with the artifact path, expected stage, user-approved digest, and …
- Very specific: line 99
Only accepted may advance, and only in this fixed order:
- Very specific: line 121
After it writes its assigned artifact, compute SHA-256 over the complete stored bytes without normalization.
≈ “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
<!-- BEGIN BLUEPRINT INTERFACES -->Generated from
blueprint.yaml. Do not edit this block by hand.
Executable Interfaces:
Send the required caller (caller skill), interface, version, and arguments; optional dry_run defaults to false. Compact uses ordered positionals plus an option mapping; ordered raw argv uses positionals: [] plus every argv token in list options. Never mix forms.
distill-to-rutters._rtx.interface.validate-and-route— Validate one source or exact-fixed-basename typed artifact and authorize no more than one deterministic route.- Caller:
distill-to-rutters - Version: 1
- Security level: 0
- Alternative:
defaultArguments JSON (replace labels with actual values). Omit optional positionals and options that are not needed. {"options": {"--approved-digest": "SHA256", "--artifact-path": "PATH", "--expected-stage": "STAGE", "--user-decision": "approve|reject"}, "positionals": [], "stdin": null} Required options: ["--approved-digest", "--artifact-path", "--expected-stage", "--user-decision"]; positional arity: 0..0; stdin: forbidden
- Caller:
Instruction Interfaces:
These are LLM-readable instruction surfaces. Read and follow them directly; do not invoke the MCP server for them.
distill-to-rutters.source.assign-rutters.interface.assign-rutters@1— Assign every and only approved part to Rutter-owned execution, preserve exact Rutter, Voyage, workflow, obligation, and coordinator identities, and require coordinator authorization for multi-Voyage policy.distill-to-rutters.source.breakdown.interface.breakdown@1— Close recursive contained behavior-defining context, include every generated projection's governing source, reject ready status for missing, unreadable, unresolved, conflicting, or changed rows, and identify behaviorally inseparable work.distill-to-rutters.source.design-implementation.interface.design-implementation@1— Design exact target changes only when the owned production public-runtime compatibility predicate passes; otherwise report the exact design block.distill-to-rutters.source.extract-evolutions.interface.extract-evolutions@1— Extract assignment-closed evolutions, Voyage membership, coordinator rules, actor-owned decisions, validators, outcomes, and a deterministic transition matrix with valid evolution or terminal targets.distill-to-rutters.source.finalize.interface.finalize@1— Finalize projections, write the exact source-stem Compass candidate, and bind its identical contained path and digest in the envelope and 07_entrypoint.md body.distill-to-rutters.source.implement.interface.implement@1— Apply exactly one approved design and report complete implemented-only trace correspondence before finalization.distill-to-rutters.source.validate-logic.interface.validate-logic@1— Validate that every normative and coordinator obligation has actor-preserving enforcement through a current versioned public Rutter binding contract.distill-to-rutters.source.verify.interface.verify@1— Verify the exact approved entrypoint predecessor and candidate identity, orchestration, and semantic preservation; any failed or blocked check prevents verified.
<!-- END BLUEPRINT INTERFACES -->Distill to Rutters
First-release scope
This release supports exactly one Rutter and one Voyage. If the approved
breakdown requires independent Voyages or coordinator policy, route to
assignment-gap and stop. Successful implementation and verification must
cover every approved logic obligation exactly once. Runtime readiness is based
on exercised public Rutter, Voyage, VoyageDispenser, and persistence behavior,
plus the declared executable Compass process-binding dependency; typed
semantic-capability declarations remain later hardening work.
Turn one Markdown instruction into a review-gated Rutter implementation. The gateway routes one stage at a time; stage interfaces own the actual analysis, authoring, and verification.
Preflight
Resolve one readable Markdown input, its owning registered skill module, and
that module's declared implementation boundary. Preserve the original file and
unrelated dirty work. The workspace is a sibling directory named from the input
stem with _distillation appended. The input, owning module, implementation
boundary, and workspace must remain inside the repository; refuse paths that
would escape it.
Every stage artifact must live in that exact <source-stem>_distillation
workspace and preserve one common root source prerequisite. Never splice an
artifact, context row, or deliverable from another source or distillation run.
Approval is bound to exact bytes. File existence or conversational assent
without a path and digest is not approval, and there is no status ledger. In a
new conversation, require the last artifact path, its gateway-reported SHA-256,
its typed outcome, and the user's explicit approve or reject decision.
Route exactly one stage
With no prior artifact, invoke
distill-to-rutters._rtx.interface.validate-and-route@1 in
source-preflight mode on the source Markdown, with an empty approved digest
and decision approve. Only its breakdown route may bootstrap the workflow;
preflight cannot authorize a later stage.
Before every post-bootstrap route, invoke that injected interface with the
artifact path, expected stage, user-approved digest, and explicit decision.
Consume only its JSON status, artifact_digest, outcome,
authorized_route, and earliest_stale_prerequisite. Do not invoke a private
filesystem implementation path.
The artifact basename must match the accepted stage exactly:
| Order | Accepted stage | Required exact filename |
| 1 | breakdown | 01_breakdown.md |
| 2 | assign-rutters | 02_rutter_assignment.md |
| 3 | extract-evolutions | 03_evolutions_and_transitions.md |
| 4 | validate-logic | 04_logic_validation.md |
| 5 | design-implementation | 05_implementation_design.md |
| 6 | implement | 06_implementation_report.md |
| 7 | finalize | 07_entrypoint.md |
| 8 | verify | 08_verification.md |
Unknown or old filenames cannot authorize routing. Treat a basename mismatch as failed approval authority even when the file's envelope, outcome, or digest would otherwise validate. A correction must use the fixed filename owned by its stage; it cannot skip or reorder a stage.
Only accepted may advance, and only in this fixed order:
| Accepted stage | Authorized next stage |
breakdown | assign-rutters |
assign-rutters | extract-evolutions |
extract-evolutions | validate-logic |
validate-logic | design-implementation |
design-implementation | implement |
implement | finalize |
finalize | verify |
verify | stop complete |
For gap or rejected, invoke only the owning stage returned as the
non-advancing repair route. For stale, preserve every file and invoke only
the earliest owning stage returned by the interface. For partial, failed,
or blocked, stop until the condition changes or the user explicitly starts
the owning stage again. A bootstrap or repair route never skips a stage and
never counts as advancement.
Pass the input path, owning module, approved prerequisite artifacts, current
user feedback, and live repository state to the selected stage interface.
After it writes its assigned artifact, compute SHA-256 over the complete stored
bytes without normalization. Report the exact (path, digest, outcome) tuple,
ask the user to validate it, and stop. The next invocation must revalidate that
tuple, every transitive prerequisite, and every present context-closure path
before routing. Validation reads each consulted file's bytes once, compares a
recorded digest before parsing a changed artifact, and rehashes all consulted
paths before accepting; concurrent mutation produces stale, never acceptance.
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.

