Commit 6d1b6cc
docs(lark-mail): add data integrity and write-confirmation rules
Add a new top-level safety section "数据真实性与操作合规" with 3 hard
constraints on agent behavior: empty result is a valid answer, do not
fabricate target objects, and explicitly confirm before destructive
write operations. Cross-reference the rule from API Resources section
to catch agents that jump directly to native API lookup.
Addresses recurring evaluation failures (c03/c04/c06/c09/c14/c19~c24/c40)
where the agent fabricated IDs or auto-executed destructive operations.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>1 parent 7fb71c6 commit 6d1b6cc
1 file changed
Lines changed: 15 additions & 0 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
40 | 40 | | |
41 | 41 | | |
42 | 42 | | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| 52 | + | |
| 53 | + | |
| 54 | + | |
| 55 | + | |
43 | 56 | | |
44 | 57 | | |
45 | 58 | | |
| |||
441 | 454 | | |
442 | 455 | | |
443 | 456 | | |
| 457 | + | |
| 458 | + | |
444 | 459 | | |
445 | 460 | | |
446 | 461 | | |
| |||
0 commit comments