knowledge-routing
Skill0 starsChanged 4 days agoMITView on GitHub
Use when you capture a learning and must decide where it lives. Route by type — shared construct to a skill, fact to memory, reference to a doc, behavior to a thin pointer — so it propagates once and is never duplicated or re-derived.
In this repository108 other files it gives its agents.
- adaptive-estimationSkill
- adaptive-loop-budgetingSkill
- adopt-and-deleteSkill
- adversarial-reviewSkill
- agent-drives-uiSkill
- agentic-coding-loopSkill
- agent-otel-trajectorySkill
- agents-as-iac-modulesSkill
- ascii-ui-mockupsSkill
- authoring-html-emailSkill
- authoring-skills-vs-mcpSkill
- batched-ssh-enumerationSkill
- c4-architectureSkill
- challenge-the-premiseSkill
- codegraph-bootstrapSkill
- codegraph-worktree-startupSkill
- colocated-pod-tolerationsSkill
- command-creatorSkill
- complexity-budgetsSkill
- connector-readinessSkill
- context-anchored-patchingSkill
- continuous-issue-resolutionSkill
- credential-proxySkill
- criticality-tiered-reportingSkill
- data-connector-buildingSkill
- declarative-firstSkill
- disk-pressure-forensicsSkill
- early-return-over-elseSkill
- email-triageSkill
- enums-codify-behaviorSkill
- ephemeral-workspace-lifecycleSkill
- event-sourcingSkill
- evidence-claim-boundariesSkill
- exhaustive-matchSkill
- feature-coverage-not-just-line-coverageSkill
- full-disk-forensicsSkill
- gate-before-pushSkill
- github-pr-shepherdingSkill
- graceful-shutdown-stateful-agentsSkill
- grounded-generationSkill
- iac-incident-responseSkill
- iac-not-ad-hocSkill
- instrumented-service-scaffoldSkill
- interface-over-conditionalsSkill
- lean-on-oss-standardsSkill
- live-money-disciplineSkill
- live-state-verificationSkill
- loop-governance-and-learningSkill
- low-level-executor-task-specSkill
- map-dispatch-over-conditionalsSkill
- merge-conflict-disciplineSkill
- mermaid-diagramsSkill
- oke-arc-runners-opentofuSkill
- orchestrate-agySkill
- orchestrate-claudeSkill
- orchestrate-codexSkill
- orchestrate-geminiSkill
- otel-emit-at-willSkill
- parameter-designSkill
- parse-dont-validateSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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 22
If a thing is two types, split it: the reusable part becomes a skill, the specific fact becomes a memory, and each points at the other.
- Distinctive: line 27
Before asking a human or re-solving a problem, check the system of record first — the home a piece of knowledge would route *to* is the …
- Distinctive: line 43
Filing a reusable construct as a one-off note: If others will reuse it, it is a skill, not a buried memory.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Knowledge Routing
Knowledge gets re-explained and re-derived for one of two reasons: it is trapped in a place only one actor can read, or it is duplicated across several places that drift apart. The fix is a single rule: capture once, route to the one right home by type, and point — never copy.
The routing rule
| What you learned | Where it lives | Why there |
| A construct / technique many actors will reuse | a skill (the single source of truth) | versioned, audited, distributed once to everyone |
| A fact / state / decision | a memory / shared store that others can read | durable and queryable; it is knowledge, not behavior |
| A reference / long-form explanation | a wiki / doc | linkable, survives in depth without bloating prompts |
| A behavior an actor must follow | a thin pointer to the skill from the config file (CLAUDE.md / AGENTS.md) | one source; the config references, never restates |
If a thing is two types, split it: the reusable part becomes a skill, the specific fact becomes a memory, and each points at the other.
Read from the same home you write to
Before asking a human or re-solving a problem, check the system of record first — the home a piece of knowledge would route to is the home you read it from. "Ask the person" is the last resort, not the first; most of what feels like a missing answer is already captured somewhere readable.
Capture once, propagate by tooling
The artifact is the source. Tooling (a bundle, a sync, a build) distributes it to every consumer. You never hand-copy the content into N files — the copy is how drift starts. Changing the behavior means changing the one source, then letting propagation carry it.
Anti-patterns
- Pasting skill content into a config file. The config should hold a pointer
("follow the
xskill"), not a second copy that will rot. - Filing a reusable construct as a one-off note. If others will reuse it, it is a skill, not a buried memory.
- A new parallel doc for something already documented. Extend the canonical one; do not fork it.
- Asking for what is already recorded. That is a read miss, not a knowledge gap — fix your lookup, not the source.
The test
For anything worth keeping, answer: "Where is the ONE place this lives, and how does everyone else read it without me telling them?" If the answer is "in my head" or "in three places," route it.
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.

