cervellaswarm/packages/lingua-universale/examples/dogfood_agent_orchestration.lu at main · rafapra3008/cervellaswarm · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
# SPDX-License-Identifier: Apache-2.0
# Lingua Universale Dogfooding: Mini Agent Orchestration
# First real program written in LU -- CervellaSwarm S454.
# 3 AI agents coordinated with formal guarantees: supervisor
# delegates analysis, worker executes, validator verifies.
# If anyone violates the protocol, the runtime blocks it.

type TaskComplexity = Simple | Moderate | Complex

type AnalysisTask =
    id: String
    description: String
    complexity: TaskComplexity

agent Supervisor:
    role: supervisor
    trust: trusted
    accepts: TaskResult
    produces: TaskRequest
    requires: task.assigned
    ensures: result.verified

agent AnalysisWorker:
    role: worker
    trust: standard
    accepts: TaskRequest
    produces: TaskResult
    requires: task.received
    ensures: analysis.complete

agent QualityValidator:
    role: validator
    trust: verified
    accepts: TaskRequest
    produces: AuditVerdict
    requires: result.submitted
    ensures: verdict.justified

protocol AgentOrchestration:
    roles: supervisor, worker, validator

    supervisor asks worker to execute analysis
    worker returns result to supervisor
    supervisor asks validator to verify result

    when validator decides:
        pass:
            validator returns approval to supervisor

        fail:
            validator sends feedback to supervisor
            supervisor tells worker revise task

    properties:
        always terminates
        no deadlock
        no deletion
        all roles participate
        trust >= standard
        task_request before task_result
        worker cannot send dm
        validator exclusive dm