code-intelligence/roadmap.json at main · oceanwap/code-intelligence · 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
{
  "version": 1,
  "updated": "2026-07-03T00-00:00Z",
  "items": [
    {
      "id": "cognition-layer",
      "title": "Cognition Layer + Superpowered Combo Tools",
      "type": "feature",
      "status": "draft",
      "prd": ".prd/prd-cognition-layer.md",
      "summary": "Lift the 25 leaf tools with a signalization layer, blackboard, composite scoring, and 4 superpowered combo tools (audit_symbol, plan_refactor, trace_workflow, collaborate) plus a goal/intent API. Leaves stay; never replaced.",
      "sprints": [
        {
          "id": "sprint-1",
          "name": "Sprint 1: P0 foundation + P1 leaf upgrades",
          "loc_estimate": 1470,
          "stories": ["US-001", "US-002"]
        },
        {
          "id": "sprint-2",
          "name": "Sprint 2: P2 composite scoring + cross-output index",
          "loc_estimate": 430,
          "stories": ["US-003"]
        },
        {
          "id": "sprint-3",
          "name": "Sprint 3: P3a audit_symbol + plan_refactor",
          "loc_estimate": 950,
          "stories": ["US-004"]
        },
        {
          "id": "sprint-4",
          "name": "Sprint 4: P3b trace_workflow + collaborate + P4 intents/recommend/session_status",
          "loc_estimate": 1700,
          "stories": ["US-005", "US-006"]
        }
      ],
      "totals": {
        "loc_estimate": 4850,
        "sprints": 4,
        "user_stories": 6
      },
      "open_questions": [
        "OQ-1: collaborate intent classification (heuristic vs Ollama)",
        "OQ-2: blast_radius storage (composite-scores.json vs attention.json)",
        "OQ-3: cross-output embedding cost (TTL/per-commit)",
        "OQ-4: recommended_next cold start (heuristic vs learned)",
        "OQ-5: audit_symbol writeToBlackboard default"
      ],
      "next_step": "Engineer for US-001 (P0 cognition foundation)"
    },
    {
      "id": "sprint8-review-pr",
      "title": "Sprint 8: review_pr combo + intelligence-layer hardening",
      "type": "feature",
      "status": "draft",
      "prd": ".prd/prd-sprint8-review-pr.md",
      "scope": "review_pr + 4 quick wins",
      "summary": "Single highest-frequency agent workflow (PR review) plus 4 architectural quick wins from the intelligence-layer review. review_pr fuses 6 leaf signals into one merge-decision call. Quick wins (B1 F5 cap propagation, B2 composite memoization, B3 strict isToolResult, B4 F6 resolver gate) close scoring/cap/discriminator gaps.",
      "sprints": [
        {
          "id": "sprint-8",
          "name": "Sprint 8: review_pr combo + B1-B4 hardening",
          "loc_estimate": 750,
          "stories": ["US-001", "US-002"]
        }
      ],
      "totals": {
        "loc_estimate": 750,
        "sprints": 1,
        "user_stories": 2
      },
      "open_questions": [
        "OQ-1: canonical merge-gate tool (review_pr vs plan_refactor vs release-prep)",
        "OQ-2: top-N cutoff on per-symbol audit_symbol calls inside review_pr (default 10)",
        "OQ-3: cross-output recall in review_pr (opt-in via env to avoid FR-11 break)",
        "OQ-4: B2 cache invalidation on graph.json changes"
      ],
      "next_step": "Engineer for US-001 (review_pr combo) + US-002 (B1-B4 quick wins)"
    }
  ]
}