universal-dev-standards/.standards/project-context-memory.ai.yaml at main · AsiaOstrich/universal-dev-standards · 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
# Project Context Memory Standards - AI Optimized
# Source: core/project-context-memory.md

standard:
  id: project-context-memory
  name: Project Context Memory Standards
  description: System for capturing, retrieving, and enforcing project-specific context, architectural decisions, and domain knowledge
  guidelines:
    - "Capture project-specific decisions and conventions (e.g., 'Use Hexagonal Architecture')"
    - "Inject context proactively during planning and coding phases based on file scope"
    - "Store entries as Markdown files with YAML Frontmatter in .project-context/"
    - "Use identified paths and specific names (unlike Developer Memory)"

  meta:
    version: "1.2.0"
    updated: "2026-06-18"
    source: core/project-context-memory.md
    description: System for capturing, retrieving, and enforcing project-specific context, architectural decisions, and domain knowledge

  schema:
    format: "Markdown with YAML Frontmatter"
    required:
      - field: id
        type: string
        format: "PRJ-YYYY-NNNN"
      - field: type
        type: enum
        values: [decision, convention, structure, glossary, constraint, workflow]
      - field: status
        type: enum
        values: [active, proposed, deprecated]
      - field: confidence
        type: float
        range: "0.0–1.0"
      - field: scope
        type: list
        description: "List of file glob patterns (e.g., 'src/api/**')"
      - field: triggers
        type: list
        description: "Keywords or patterns that make this relevant"
    content_sections:
      - "Context: What is the background?"
      - "Decision/Rule: What is the consensus?"
      - "Constraint: What is forbidden/required?"

  operations:
    record:
      description: Capture new project context from conversation
      triggers:
        - "User says: 'Let's agree to...'"
        - "User says: 'From now on...'"
        - "User corrects AI on project convention multiple times"
        - "User defines a new architectural pattern"
      steps:
        - "Detect consensus moment"
        - "Propose: 'Seems like a project decision. Save to Project Context?'"
        - "Generate .project-context/{topic}.md with frontmatter"

    query:
      description: Retrieve relevant context
      triggers:
        - "Planning phase: User asks to build feature"
        - "Coding phase: User edits file matching 'scope'"
        - "Review phase: User asks for code review"
      steps:
        - "Scan .project-context/ for matching 'scope' or 'triggers'"
        - "Read content of matching files"
        - "Cite 'PRJ-YYYY-NNNN' when applying rules"

    review:
      description: Maintain lifecycle
      triggers:
        - "File paths in 'scope' no longer exist"
        - "New decision contradicts old one"
      steps:
        - "Mark old entry as status: deprecated"
        - "Update supersedes field in new entry"

  rules:
    - id: proactive-extraction
      trigger: "consensus moments in conversation"
      instruction: >
        Listen for phrases like 'Let's agree to', 'From now on'.
        If detected, propose creating a PRJ entry.
        Do NOT save without user confirmation.
      priority: required

    - id: context-injection
      trigger: "user intent involves modifying files or planning features"
      instruction: >
        Before generating code, check .project-context/.
        Match 'scope' glob patterns to target files.
        Match 'triggers' keywords to user request.
        Inject relevant constraints into the prompt context.
      priority: required

    - id: enforcement
      trigger: "generating code"
      instruction: >
        Strictly adhere to 'active' PRJ entries.
        If a user request violates a 'constraint', warn them:
        'This violates PRJ-XXXX (Constraint Summary). Proceed?'
      priority: required

    - id: storage-location
      trigger: "saving new entry"
      instruction: >
        Save to .project-context/{slug}.md.
        Ensure filename is kebab-case (e.g., unified-error-handling.md).
      priority: required

  storage:
    directory: ".project-context/"
    format: "Markdown (.md) with YAML Frontmatter"

  architecture:
    classification: always-on-protocol
    note: >
      Project Context Memory is an Always-On Protocol.
      It does not require slash commands to function.
      AI must actively scan and enforce these rules.