converge-plan
Skill0 starsChanged 2 months agoMITView on GitHub
Stress-test a costly or hard-to-reverse plan before implementation by freezing a rubric, iterating improvements and adversarial critique, and ending with one crux and the cheapest falsifiable experiment. Use for “converge this plan”, “수렴”, major architecture choices, or risky research bets—not routine edits.
In this repository9 other files it gives its agents.
- absorbSkill
- agent-starterSkill
- aim-before-buildSkill
- decision-sheetSkill
- evidence-auditSkill
- goal-contractSkill
- harness-auditSkill
- morning-briefSkill
- phase-loopSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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 8
Use this before implementation, not as a generic review after code exists.
- Distinctive: line 14
Stop after two consecutive rounds with no new surviving defect, two rounds with less than 5% improvement, or a hard cap of 6–8 rounds.
- Distinctive: line 15
Do not declare convergence without at least one primary-source comparison and one direct check of the real system when those are available.
- Distinctive: line 23
Stop reason and remaining uncertainty
- Distinctive: line 25
Do not change the rubric mid-run or accept an improvement without trying to disprove it.
≈ “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
Converge Plan
Use this before implementation, not as a generic review after code exists.
Workflow
- Freeze a small rubric before round one. Suggested dimensions: feasibility, evidence, gap coverage, differentiation, cost realism, and falsifiability.
- For each round: improve the largest weakness, score with evidence, attempt to refute the improvement, then record surviving new defects.
- Stop after two consecutive rounds with no new surviving defect, two rounds with less than 5% improvement, or a hard cap of 6–8 rounds.
- Do not declare convergence without at least one primary-source comparison and one direct check of the real system when those are available.
Output
- Frozen rubric and score trajectory
- Accepted plan changes and rejected alternatives
- One irreversible crux
- One cheapest falsifiable experiment
- Stop reason and remaining uncertainty
Do not change the rubric mid-run or accept an improvement without trying to disprove it.
Next
Turn the converged outcome into goal-contract. If the crux is unresolved, run only the cheapest experiment first.
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.