structure-discovery
Skill8 starsChanged 57 days agoMITView on GitHub
Find the hidden pattern that organizes the mess. Use for "these look related but we can't say how", classification with suspicious gaps, reverse-engineering an undocumented system or format, "is there a symmetry we're not using?", scale-free or fat-tailed data, or generating conjectures from computed special cases.
In this repository16 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- boundary-designSkill
- causal-auditSkill
- decision-bias-checkSkill
- estimationSkill
- evidence-synthesisSkill
- experience-and-transmissionSkill
- failure-forensicsSkill
- formal-correctnessSkill
- measurement-disciplineSkill
- narrative-sensemakingSkill
- normative-designSkill
- problem-reframingSkill
- representation-and-possibilitySkill
- systems-leverageSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 45
If no shape above matches, use a built-in agent with the closest GENIUS: pattern from rules/agent-routes.json instead.
≈ “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
Structure Discovery
Problem shape: many observations, one suspected hidden regularity — an unnamed axis, an unexploited symmetry, an undeciphered format, a topology forced by constraints. The move: make the structure explicit and falsifiable, then let it predict something you haven't seen yet.
Relevant geniuses
| Agent | Use when |
| mendeleev | many known items with a suspected hidden ordering — tabulate, leave explicit gaps, predict the gaps' properties |
| noether | hidden regularity via invariance — find the symmetry group, quotient the search space, treat symmetry breaking as signal |
| kekule | components with known connection constraints, unknown structure — count the bonds and let the count force the shape |
| vonneumann | the problem looks isomorphic to a solved one in another field — find the mapping, import the solution |
| rejewski | black-box system to reconstruct from outputs — structural invariants, exploit the procedure around the algorithm |
| champollion | undeciphered format with a partial parallel (bilingual) sample — anchor known fragments and propagate |
| ventris | no parallel text exists — grid the internal regularities, then test by prediction |
| poincare | qualitative behavior before quantitative solution — topological equivalence, structural stability |
| mandelbrot | pattern repeats across scales; distribution has fat tails being treated as mild randomness |
| ramanujan | need many candidate patterns fast — compute 50+ special cases, conjecture, then mandatory prover handoff (never ship unverified) |
Invocation
- Pick the best-fit agent above. If two or more fit, run
tools/genius-invoker.sh route "<problem>"and take the top ranked match. - Load it:
tools/genius-invoker.sh invoke <agent> "<problem>", then readagents/genius/<agent>.mdin full. - Apply the agent's
<workflow>step by step and answer in its<output-format>. A discovered structure must predict something checkable — a gap's properties, a next case, an invariant — or it is decoration. - ramanujan output is conjecture by definition: chain a prover
(
tools/genius-invoker.sh compose ramanujan lamport -- "<problem>"or hand to dijkstra) before anything ships. - If no shape above matches, use a built-in agent with the closest
GENIUS:pattern fromrules/agent-routes.jsoninstead.
Refuse when
- The pattern is wanted for a narrative, with no willingness to test its predictions.
- The data is too sparse to distinguish structure from apophenia — say so.
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.