Forge
markdowna5f71db5
1# 1. Ask the model
2
3Most teams work in "instruction-first" mode.
4The agent-oriented approach (Agent-first) reverses the order: "diagnose the deficit first, then act."
5
6```text
7Ordinary mode:
8human -> instruction -> agent executes
9
10Agent-oriented mode:
11human -> question about deficit -> environment/tools -> agent executes
12```
13
14## Why it matters
15
16When the agent lacks context, memory, or tools, it starts guessing.
17From the outside this looks like "the model is dumb." In reality it's often "the environment forces guessing."
18
19## Concrete example
20
21"I need diagnostics by position, not log parsing" — you give the agent `roslyn_get_diagnostics` instead of forcing it to guess from build output.
22"I lose the thread after summarization" — you give it notes it controls and a protocol: user reminds, agent reads notes.
23
24In both cases the deficit is resolved not by rephrasing the prompt, but by changing the environment.
25
26## Basic questions to the agent
27
28Before a task:
29- What do you need to solve this well?
30- What's missing right now: data, tools, time, access?
31- Where is the risk of error highest?
32
33After a task:
34- What worked?
35- What was the bottleneck?
36- What should we add to the process so it's better next time?
37
38## Minimal protocol
39
401. State the goal in one sentence.
412. Ask the agent about its deficits.
423. Close critical deficits (context / tool / rule).
434. Run a short execution cycle.
445. Capture takeaways in notes.
45
46## Success criterion
47
48Success is not "the agent guessed right."
49Success is when the agent **explains** why it's asking for a specific resource, and after getting it quality actually improves.
50
View only · write via MCP/CIDE