experiment-catalog
AGENTS.md8 starsChanged 5 months agoMITView on GitHub
Run from repository root:
- 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 31,439 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.
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
17 rules. About 0 are common, 11 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
6
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
- Distinctive: line 12
Run from repository root:
- Distinctive: line 25
Prefer deterministic scripts over interactive/manual steps.
- Very specific: line 26
Keep command names stable (smoke, check, test, ci).
- Distinctive: line 27
Update docs and scripts in the same change when workflow behavior changes.
- Distinctive: line 28
Avoid side effects outside the repo unless explicitly required.
- Very specific: line 31
Do not commit .env files; use .env.example patterns for configuration.
- Very specific: line 47
Keep internal data models typed and normalized (see catalog/models/).
- Distinctive: line 48
Keep each module focused on one responsibility.
- Very specific: line 49
Document boundary ownership in docs/ARCHITECTURE.md.
- Distinctive: line 56
Keep event fields stable for querying and alerting.
- Distinctive: line 63
Update status checkpoints during execution and after major decisions.
- Very specific: line 67
Run make check before make test.
- Very specific: line 68
Run make ci before pushing large refactors.
- Distinctive: line 69
Treat lint/type failures as blocking.
- Distinctive: line 73
Remove stale scripts/docs quickly.
- Distinctive: line 74
Keep templates and real workflows in sync.
- Distinctive: line 75
Run periodic harness audits:
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 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
Project Overview
- Project:
experiment-catalog - Primary runtime(s): .NET 10 (C#), Node/Svelte (TypeScript), Python
- Main entrypoint(s):
catalog/Program.cs(API),evaluator/Program.cs(evaluator),ui/(Svelte SPA) - Solution file:
experiment-catalog.sln
Harness Commands
Run from repository root:
| Goal | Command |
| Dev environment setup | make setup |
| Fast sanity check | make smoke |
| Static checks | make check |
| Full test suite | make test |
| Security scanning | make security |
| CI-equivalent local run | make ci |
Constraints And Guardrails
- Prefer deterministic scripts over interactive/manual steps.
- Keep command names stable (
smoke,check,test,ci). - Update docs and scripts in the same change when workflow behavior changes.
- Avoid side effects outside the repo unless explicitly required.
- The .NET solution requires
dotnetSDK 10.0+. - The UI requires Node.js 20+ and npm.
- Do not commit
.envfiles; use.env.examplepatterns for configuration.
Project Structure
| Path | Description |
catalog/ | ASP.NET Core API service (exp-catalog) with MCP server, REST controllers, and Azure Blob storage |
catalog.tests/ | xUnit test project for catalog |
evaluator/ | ASP.NET Core evaluator service |
evaluation/ | Python-based evaluation scripts |
ui/ | Svelte 5 SPA with Vite |
catalog.Dockerfile | Container image for catalog + UI |
Architecture Boundaries
- Parse and validate external data at boundaries (controllers, MCP tools).
- Keep internal data models typed and normalized (see
catalog/models/). - Keep each module focused on one responsibility.
- Document boundary ownership in
docs/ARCHITECTURE.md.
Observability Expectations
- OpenTelemetry is already instrumented in catalog and evaluator via Azure Monitor exporter.
- Include
trace_idandrun_idin long-running workflow logs. - Emit structured event names for major transitions (start, step, success, failure).
- Keep event fields stable for querying and alerting.
- Maintain field definitions in
docs/OBSERVABILITY.md.
Execution Plans
- For tasks expected to exceed ~30 minutes, create/update
PLANS.mdbefore coding. - Track scope, constraints, milestones, and verification steps.
- Update status checkpoints during execution and after major decisions.
Static Analysis And Quality Gates
- Run
make checkbeforemake test. - Run
make cibefore pushing large refactors. - Treat lint/type failures as blocking.
Entropy Management
- Remove stale scripts/docs quickly.
- Keep templates and real workflows in sync.
- Run periodic harness audits:
scripts/audit_harness.sh .
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.