Spaces:
Running
Running
SilverElixir commited on
Commit ·
fc369dc
1
Parent(s): 92ad8f0
AGENTS.md: use context7 MCP tools for current library docs
Browse files
AGENTS.md
CHANGED
|
@@ -37,6 +37,9 @@
|
|
| 37 |
- Comments explain why, not what. Keep the dated evidence comments that prevent regressions, but keep them short and move long write-ups to docs/.
|
| 38 |
- No filler prose, emojis, or heavy dash use in code, comments, or docs.
|
| 39 |
|
|
|
|
|
|
|
|
|
|
| 40 |
## Session hygiene (the owner does not track this, guide them)
|
| 41 |
- Effort: for tasks above the default (High: streaming, security, refactor; XHigh: audits, architecture, production incidents), say which level fits and why, then wait for the owner's OK. For default-level tasks (Low: read, explain, rename; Medium: single-module fixes, tests, docs) just proceed. You cannot change effort yourself; the owner does.
|
| 42 |
- Suggest a new chat when the topic changes completely, and /compact when the session gets long.
|
|
|
|
| 37 |
- Comments explain why, not what. Keep the dated evidence comments that prevent regressions, but keep them short and move long write-ups to docs/.
|
| 38 |
- No filler prose, emojis, or heavy dash use in code, comments, or docs.
|
| 39 |
|
| 40 |
+
## Docs lookup
|
| 41 |
+
- When you need current library docs (aiogram, FastAPI, google-genai, Deno) rather than memory, use the `context7` MCP tools.
|
| 42 |
+
|
| 43 |
## Session hygiene (the owner does not track this, guide them)
|
| 44 |
- Effort: for tasks above the default (High: streaming, security, refactor; XHigh: audits, architecture, production incidents), say which level fits and why, then wait for the owner's OK. For default-level tasks (Low: read, explain, rename; Medium: single-module fixes, tests, docs) just proceed. You cannot change effort yourself; the owner does.
|
| 45 |
- Suggest a new chat when the topic changes completely, and /compact when the session gets long.
|