nexus-kb
AGENTS.md1 starsChanged 5 months agoMITView on GitHub
This file defines the operating baseline for AI coding agents and human contributors working in this repository. The repository contains:
- Reads credentials
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 repository5 other files it gives its agents.
- .claude/CLAUDE.mdCLAUDE.md
- mcp-maintainerSkill
- release-guardSkill
- repo-onboardingSkill
- deploySkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
16 rules. About 1 is common, 12 are distinctive, and 3 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈12
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
- could be automatic: a setting could enforce it
- Very specific: line 15
Read README.md, docs/architecture.md, and docs/phase-checklist.md before changing architecture, rules, or implementation structure.
- Distinctive: line 16
Check the target area before editing:
- Distinctive: line 20
Keep changes scoped. Do not rewrite unrelated generated files, local runtime state, or private configuration.
- Distinctive: line 21
Prefer explicit, reproducible commands.
- Distinctive: line 22
Before writing code, inspect existing patterns and update tests or verification steps when behavior changes.
- Very specific: line 26
Treat docs/architecture.md as the source of truth.
- Distinctive: line 27
Use clean boundaries between UI, backend services, workers, MCP connectors, and storage.
- Distinctive: line 28
Do not bypass RBAC, audit logging, or human review in workflows that process sensitive documents.
- Distinctive: line 29
Keep LLM calls behind a gateway that supports model routing, caching, retries, and observability.
- Distinctive: line 30
Keep raw documents, model files, vector snapshots, and audit exports out of git.
- Common: line 34
Never commit credentials, API keys, internal hostnames, customer data, logs, local sessions, or agent memory.
- Very specific: line 35
Use .env.example for required configuration keys and document expected values without secrets.
- Distinctive: line 36
Prefer least-privilege access for MCP connectors and storage clients.
- Distinctive: line 49
Avoid marketing claims that are not backed by implementation or documentation.
- Distinctive: line 50
Use consistent terminology: RAG, Knowledge Graph, MCP, RBAC, audit, queue, worker, vector database.
- Distinctive: line 51
Keep examples public-safe and free of company-private data.
≈ “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
Repository Agent Guide
This file defines the operating baseline for AI coding agents and human contributors working in this repository.
Scope
The repository contains:
- Architecture documentation for Enterprise Knowledge Hub.
- Project-scoped agent configuration under
.codex/,.claude/,.agents/, and.hermes/. - A runnable Qdrant multi-node demo under
qdrant-multi-node-cluster/.
Required Workflow
- Read
README.md,docs/architecture.md, anddocs/phase-checklist.mdbefore changing architecture, rules, or implementation structure. - Check the target area before editing:
- Root and
docs/: project documentation and governance. .codex/,.claude/,.agents/,.hermes/: agent configuration and local automation.qdrant-multi-node-cluster/: runnable Python and Docker demo.
- Root and
- Keep changes scoped. Do not rewrite unrelated generated files, local runtime state, or private configuration.
- Prefer explicit, reproducible commands. Do not rely on undocumented local state.
- Before writing code, inspect existing patterns and update tests or verification steps when behavior changes.
Architecture Rules
- Treat
docs/architecture.mdas the source of truth. - Use clean boundaries between UI, backend services, workers, MCP connectors, and storage.
- Do not bypass RBAC, audit logging, or human review in workflows that process sensitive documents.
- Keep LLM calls behind a gateway that supports model routing, caching, retries, and observability.
- Keep raw documents, model files, vector snapshots, and audit exports out of git.
Security Rules
- Never commit credentials, API keys, internal hostnames, customer data, logs, local sessions, or agent memory.
- Use
.env.examplefor required configuration keys and document expected values without secrets. - Prefer least-privilege access for MCP connectors and storage clients.
- Fail closed for authorization checks and document access checks.
Verification Guidance
- Documentation changes: review links and headings manually.
- Qdrant demo changes: run tests from
qdrant-multi-node-cluster/withmake testorpython -m unittest discover -s tests. - Agent configuration changes: verify that referenced files exist and local-only files remain ignored.
- Phase changes: use
docs/phase-checklist.mdas the completion checklist and do not mark a phase complete without the listed tests or equivalent evidence.
Writing Standards
- Be direct and technical.
- Avoid marketing claims that are not backed by implementation or documentation.
- Use consistent terminology: RAG, Knowledge Graph, MCP, RBAC, audit, queue, worker, vector database.
- Keep examples public-safe and free of company-private data.
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.

