agent-genome-lab
AGENTS.md2 starsChanged 4 months agoMITView on GitHub
This repository uses the NVE Universal Failure Genome + SkillGraph superstructure (v2.3.0).
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository24 other files it gives its agents.
- Academic Paper SummarizationSkill
- Accessibility WCAG AuditSkill
- API Error TroubleshootingSkill
- CI/CD Pipeline TroubleshootingSkill
- Competitor Analysis SynthesisSkill
- Customer Onboarding AssistanceSkill
- Dependency Audit ResolutionSkill
- Docker Image OptimizationSkill
- GDPR Data MappingSkill
- genome-analyzerSkill
- incident-distillerSkill
- Kubernetes Pod DebuggingSkill
- nve-genome-explorerSkill
- OWASP Input ValidationSkill
- repo-auditorSkill
- research-packagerSkill
- rule-patcherSkill
- Secret Leak RemediationSkill
- skill-curatorSkill
- skill-package-builderSkill
- SOC2 Policy VerificationSkill
- tama-community-summarizerSkill
- Technology Stack EvaluationSkill
- Ticket Triage and RoutingSkill
The file
We split this AGENTS.md into the rules it gives an agent
10 rules. About 1 is common, 6 are distinctive, and 3 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈6
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
- Distinctive: line 6
Before changing architecture, memory, workflows, sharing, or Failure Genome behavior, read:
- Distinctive: line 22
Do not delete valuable existing assets without explicit approval.
- Distinctive: line 23
Use docs as the system of record.
- Very specific: line 24
Keep canonical memory structured in .evolution/.
- Distinctive: line 25
Prefer small verifiable changes over sweeping rewrites.
- Distinctive: line 29
Only ask the user when the action is destructive, security-sensitive, or schema-breaking.
- Common: line 30
Never share raw code, secrets, or full logs by default.
- Distinctive: line 32
Before every commit: check "were there uncaptured incidents since last commit?"
- Very specific: line 33
Skill layer: follow .agents/rules/35-skill-admission-and-quarantine.md for skill governance.
- Very specific: line 34
Metadata-first: search skill INDEX before loading full skill content (see .agents/workflows/35-skill-capture-and-admission.md).
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 AGENTS.md 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
AGENTS.md
This repository uses the NVE Universal Failure Genome + SkillGraph superstructure (v2.3.0).
Read first
Before changing architecture, memory, workflows, sharing, or Failure Genome behavior, read:
README.mddocs/UNIVERSAL_ARCHITECTURE.mddocs/INSTALL_INTO_ANTIGRAVITY.mddocs/SAFE_MIGRATION_PLAYBOOK.mddocs/RESEARCH_FOUNDATIONS.mddocs/USER_FLOW_AUDIT_AND_SHARE.mddocs/FAILURE_GENOME_HYPOTHESIS_V1.mddocs/FAILURE_GENOME_EXPERIMENT_PLAN_V1.mddocs/OLD_TO_NEW_MAP.mddocs/ANTIGRAVITY_PROMPT_SEQUENCE.mddocs/SKILLGRAPH_UPGRADE_ARCHITECTURE.mddocs/SKILLGRAPH_OPERATING_GUIDE.mddocs/SKILLNET_GAP_MAP.md
Operating contract
- Do not delete valuable existing assets without explicit approval.
- Use docs as the system of record. This file is only a table of contents.
- Keep canonical memory structured in
.evolution/. - Prefer small verifiable changes over sweeping rewrites.
- Read
.evolution/MEMORY.mdat session start — it contains verified lessons and anti-patterns. - For each significant failure or prevention event, create an incident event immediately (see
.agents/rules/25-incident-auto-capture.mdfor trigger events). - For repeated or high-value failures, create or update a Failure Genome.
- Only ask the user when the action is destructive, security-sensitive, or schema-breaking.
- Never share raw code, secrets, or full logs by default.
- Legacy paths may remain as wrappers until migration is stable.
- Before every commit: check "were there uncaptured incidents since last commit?"
- Skill layer: follow
.agents/rules/35-skill-admission-and-quarantine.mdfor skill governance. - Metadata-first: search skill INDEX before loading full skill content (see
.agents/workflows/35-skill-capture-and-admission.md).
Canonical homes
.agents/rules/— always-on constraints.agents/workflows/— repeatable routines.agents/skills/— reusable specialized knowledge.evolution/incidents/— canonical incident events.evolution/experience_units/— distilled reusable knowledge.evolution/failure_genomes/— family-level learning units.evolution/skills/— canonical skill registry (candidate/admitted/quarantined/rejected).evolution/skill_packages/— task-oriented skill bundles.evolution/skill_relations/— semantic and dependency graph.evolution/audits/— local audit outputs.evolution/manifests/— repo manifests.evolution/exports/— shareable research packs
Done criteria
Architecture work is done only when:
- docs are updated,
- the structure exists,
- schemas match payloads,
- legacy value remains discoverable,
- and a future agent can understand the repo quickly.
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.

