forgetful
AGENTS.md341 starsChanged 23 days agoMITView on GitHub
Before opening issues or pull requests, read the contributor guide's AI tool guidance and submission policy. Before changing code or reasoning about this repository's architecture, read the architecture overview. Consult the detailed class map only as needed, using the overview's reading advisories. Do not load the entire map into context by default. After major code changes, check whether the architecture diagram or explanations are still accurate and update them if needed. Small changes do not require diagram regeneration. Preserve reviewed…
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 repository11 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- forgetful-cli-setupSkill
- forgetful-context-gatherSkill
- forgetful-encode-repoSkill
- forgetful-entitiesSkill
- forgetful-exploreSkill
- forgetful-filesSkill
- forgetful-mcp-setupSkill
- forgetful-proceduresSkill
- forgetful-recallSkill
- forgetful-rememberSkill
The file
We split this AGENTS.md into the rules it gives an agent
9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
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
- Distinctive: line 5
Before opening issues or pull requests, read the contributor guide's AI tool guidance and submission policy.
- Distinctive: line 11
Before changing code or reasoning about this repository's architecture, read the architecture overview.
- Distinctive: line 17
After major code changes, check whether the architecture diagram or explanations are still accurate and update them if needed.
- Distinctive: line 36
No pollution of service layer with integration/implementation details
- Very specific: line 38
Run uv run pytest tests/architecture/ to check and prove the agreed import boundaries.
- Distinctive: line 47
Purpose: Test business logic with stubbed I/O (no real database required)
- Distinctive: line 58
Purpose: Test complete stack with in-memory SQLite
- Distinctive: line 68
Purpose: Test complete stack with real PostgreSQL
- Distinctive: line 79
Ensure that you run ruff following any changes and address any issues raised
≈ “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
Forgetful Development Guide
Contributions
Before opening issues or pull requests, read the contributor guide's AI tool guidance and submission policy.
Architecture
Before changing code or reasoning about this repository's architecture, read the architecture overview.
Consult the detailed class map only as needed, using the overview's reading advisories. Do not load the entire map into context by default.
After major code changes, check whether the architecture diagram or explanations are still accurate and update them if needed. Small changes do not require diagram regeneration. Preserve reviewed explanations that remain accurate. Keep the overview at most 500 lines; it has no column-width limit, including Mermaid statements.
Mycelium is optional tooling, not a prerequisite for contributing or updating the overview.
Its architecture skill and map-generation skill
are available in ScottRBK/mycelium if useful. If refreshing the generated backup,
use fresh analysis, keep the JSON outside the repository, and export full detail with
--max-classes 1000, --test-path tests, and --test-path test_harness/runs.
Layered architecture: routes -> services -> protocols -> repositories/adapters No pollution of service layer with integration/implementation details
Run uv run pytest tests/architecture/ to check and prove the agreed import boundaries.
Testing Philosophy
We focus on integration and E2E tests over unit tests. Tests should cover critical workflows without exhaustive edge case coverage.
Aim for each test to finish in under 15 seconds; this is advisory, with no enforced timing checks.
Integration Tests
Location: tests/integration/
Purpose: Test business logic with stubbed I/O (no real database required)
Run locally:
uv run pytest tests/integration/These tests use in-memory stubs and run fast (~seconds). They form the bulk of our test suite and catch 90% of issues.
End-to-End Tests
SQLite E2E Tests
Location: tests/e2e_sqlite/
Purpose: Test complete stack with in-memory SQLite
Requirements: None (no Docker required)
Run locally:
uv run pytest tests/e2e_sqlite/These tests use an in-memory SQLite database for test isolation. Fast execution with automatic cleanup.
PostgreSQL E2E Tests
Location: tests/e2e/
Purpose: Test complete stack with real PostgreSQL
Requirements: PostgreSQL running in Docker
Run locally:
cd docker && docker compose down -v
uv run pytest -m e2eRemember: rebuild docker image if running local container service, not required however for e2e tests.
Linting
Ensure that you run ruff following any changes and address any issues raised
uv tool run ruff check .Note: Ruff UP006 rule enforces Python 3.12+ built-in generics (list instead of typing.List, dict instead of typing.Dict, etc.). This catches legacy type hint syntax automatically.
Skills
The solution has a catalogue of skills that is used to help drive agent behaviour, whenever adding or modifying behaviour around tools or the system, you should look to review these and consider whether these need updating as part of the changes you have made to the solution as well.
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.

