devspark/full-cycle.harness.yaml at main · markhazleton/devspark · 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
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
apiVersion: devspark.ai/v1
kind: HarnessSpec
name: full-cycle

# Genuinely unattended lifecycle runner. Unlike templates/workflows/full-cycle.yaml
# (which is sequenced by `devspark run` and expects an already-present driving
# agent to read each "invoke X via Y" line), this spec is executed by
# `devspark harness run`, which shells out to a real adapter CLI per step.
#
#   devspark harness run full-cycle.harness.yaml --adapter claude_code --hands-off
#
# Run `devspark adapter doctor` first to confirm a write-capable adapter is
# actually available (claude_code/codex shell out to `claude`/`codex --print`;
# copilot/cursor require their respective CLIs on PATH).
#
# Validation rules below deliberately avoid hardcoding a feature directory —
# none exists yet when this spec is authored — by resolving the most recently
# modified .documentation/specs/*/ directory at validation time instead.

scope:
  type: repo

defaults:
  adapter: noop
  mode: agent
  retry:
    maxAttempts: 2
    backoff: fixed

telemetry:
  emit_jsonl: true
  run_dir: .documentation/devspark/runs

steps:
  - id: specify
    name: Write Feature Specification
    type: agent_task
    prompt_file: templates/commands/specify.md
    validation:
      - id: spec-present
        type: command.exit_code
        severity: error
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)spec.md\""
    on_success: plan

  - id: plan
    name: Build Implementation Plan
    type: agent_task
    prompt_file: templates/commands/plan.md
    validation:
      - id: plan-present
        type: command.exit_code
        severity: error
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)plan.md\""
    on_success: generate-tasks

  - id: generate-tasks
    name: Generate Task List
    type: agent_task
    prompt_file: templates/commands/tasks.md
    validation:
      - id: tasks-present
        type: command.exit_code
        severity: error
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)tasks.md\""
    on_success: critic

  - id: critic
    name: Adversarial Risk Review
    type: agent_task
    prompt_file: templates/commands/critic.md
    validation:
      - id: critic-gate-present
        type: command.exit_code
        severity: warning
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)gates/critic.md\""
    on_success: analyze

  - id: analyze
    name: Cross-Artifact Consistency Review
    type: agent_task
    prompt_file: templates/commands/analyze.md
    validation:
      - id: analyze-gate-present
        type: command.exit_code
        severity: warning
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)gates/analyze.md\""
    on_success: remediate-gates

  - id: remediate-gates
    name: Merge Gate Findings Into Tasks
    type: agent_task
    prompt_file: templates/commands/tasks.md
    validation:
      - id: tasks-still-present
        type: command.exit_code
        severity: warning
        command: "test -f \"$(ls -td .documentation/specs/*/ 2>/dev/null | head -1)tasks.md\""
    on_success: implement

  - id: implement
    name: Execute Tasks
    type: agent_task
    prompt_file: templates/commands/implement.md
    validation:
      - id: delivery-evidence
        type: command.exit_code
        severity: error
        command: "git diff --name-only origin/main...HEAD -- src/ test/ | grep -q ."
    on_success: pre-pr-human-review

  - id: pre-pr-human-review
    name: Human Review Gate
    type: human_gate
    mode: manual
    adapter: manual
    prompt_file: templates/commands/create-pr.md
    on_success: create-pr

  - id: create-pr
    name: Draft or Update Pull Request
    type: agent_task
    prompt_file: templates/commands/create-pr.md
    validation:
      - id: pr-exists
        type: command.exit_code
        severity: warning
        command: "gh pr view --json number >/dev/null 2>&1"
    on_success: review-pr

  - id: review-pr
    name: Constitution-Based PR Review
    type: agent_task
    prompt_file: templates/commands/pr-review.md
    validation:
      - id: review-dir-present
        type: command.exit_code
        severity: warning
        command: "test -d .documentation/specs/pr-review"