{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://awesomecua.github.io/schemas/github-candidate-observation-manifest.schema.json",
  "title": "AwesomeCUA public candidate observation manifest",
  "description": "The observation population and source ledger are independent of the active catalog comparison context.",
  "type": "object",
  "required": [
    "schema_version",
    "provider",
    "scope",
    "observation_snapshot_sha256",
    "source_ledger_sha256",
    "input_fingerprints",
    "scan",
    "active_catalog_context",
    "count",
    "supporting_primary_bodies",
    "observed_from",
    "observed_through",
    "review_policy",
    "count_policy",
    "privacy",
    "export",
    "schema",
    "manifest_schema"
  ],
  "properties": {
    "schema_version": {
      "type": "integer",
      "const": 1
    },
    "provider": {
      "const": "github"
    },
    "scope": {
      "const": "candidate-observations"
    },
    "observation_snapshot_sha256": {
      "type": "string",
      "pattern": "^[0-9a-f]{64}$"
    },
    "source_ledger_sha256": {
      "type": "string",
      "pattern": "^[0-9a-f]{64}$"
    },
    "input_fingerprints": {
      "type": "array",
      "items": {
        "type": "object",
        "required": [
          "id",
          "files"
        ],
        "properties": {
          "id": {
            "type": "string",
            "pattern": "^[0-9a-f]{64}$"
          },
          "files": {
            "type": "object",
            "required": [
              "catalog.jsonl",
              "candidates.jsonl",
              "primary-evidence.jsonl",
              "repository-aliases.json",
              "candidate-reviews.json"
            ],
            "properties": {
              "catalog.jsonl": {
                "type": "object",
                "required": [
                  "bytes",
                  "sha256"
                ],
                "properties": {
                  "bytes": {
                    "type": "integer",
                    "minimum": 0
                  },
                  "sha256": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              },
              "candidates.jsonl": {
                "type": "object",
                "required": [
                  "bytes",
                  "sha256"
                ],
                "properties": {
                  "bytes": {
                    "type": "integer",
                    "minimum": 0
                  },
                  "sha256": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              },
              "primary-evidence.jsonl": {
                "type": "object",
                "required": [
                  "bytes",
                  "sha256"
                ],
                "properties": {
                  "bytes": {
                    "type": "integer",
                    "minimum": 0
                  },
                  "sha256": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              },
              "repository-aliases.json": {
                "type": "object",
                "required": [
                  "bytes",
                  "sha256"
                ],
                "properties": {
                  "bytes": {
                    "type": "integer",
                    "minimum": 0
                  },
                  "sha256": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              },
              "candidate-reviews.json": {
                "type": "object",
                "required": [
                  "bytes",
                  "sha256"
                ],
                "properties": {
                  "bytes": {
                    "type": "integer",
                    "minimum": 0
                  },
                  "sha256": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              }
            },
            "additionalProperties": false
          }
        },
        "additionalProperties": false
      },
      "minItems": 1,
      "uniqueItems": true
    },
    "scan": {
      "type": "object",
      "required": [
        "configuration",
        "runs",
        "queries",
        "request_outcomes",
        "global_completeness_established",
        "limit"
      ],
      "properties": {
        "configuration": {
          "type": "object",
          "required": [
            "topics",
            "created_after",
            "created_before"
          ],
          "properties": {
            "topics": {
              "type": "array",
              "items": {
                "enum": [
                  "android-automation",
                  "browser-agent",
                  "browser-agents",
                  "browser-automation",
                  "computer-use",
                  "desktop-automation",
                  "gui-agent",
                  "gui-agents",
                  "gui-automation",
                  "computer-use-agent",
                  "computer-use-agents",
                  "gui-grounding",
                  "browser-use",
                  "web-agent",
                  "web-agents",
                  "desktop-agent",
                  "mobile-agent"
                ]
              },
              "minItems": 1,
              "uniqueItems": true
            },
            "created_after": {
              "type": "string",
              "format": "date"
            },
            "created_before": {
              "type": "string",
              "format": "date"
            }
          },
          "additionalProperties": false
        },
        "runs": {
          "type": "array",
          "items": {
            "type": "object",
            "required": [
              "id",
              "input_id",
              "started_at",
              "finished_at",
              "status",
              "requests_attempted",
              "before_unique_repositories",
              "new_unique_repositories",
              "accepted_metadata_rows",
              "after_unique_repositories"
            ],
            "properties": {
              "id": {
                "type": "string",
                "pattern": "^run:[0-9a-f]{24}$"
              },
              "input_id": {
                "type": "string",
                "pattern": "^[0-9a-f]{64}$"
              },
              "started_at": {
                "type": "string",
                "format": "date-time"
              },
              "finished_at": {
                "type": "string",
                "format": "date-time"
              },
              "status": {
                "enum": [
                  "rate_limited",
                  "wall_budget",
                  "request_budget",
                  "partial",
                  "finished"
                ]
              },
              "requests_attempted": {
                "type": "integer",
                "minimum": 0
              },
              "before_unique_repositories": {
                "type": "integer",
                "minimum": 0
              },
              "new_unique_repositories": {
                "type": "integer",
                "minimum": 0
              },
              "accepted_metadata_rows": {
                "type": "integer",
                "minimum": 0
              },
              "after_unique_repositories": {
                "type": "integer",
                "minimum": 0
              }
            },
            "additionalProperties": false
          }
        },
        "queries": {
          "type": "array",
          "items": {
            "oneOf": [
              {
                "type": "object",
                "required": [
                  "id",
                  "kind",
                  "status",
                  "topic",
                  "created_after",
                  "created_before",
                  "reported_total"
                ],
                "properties": {
                  "id": {
                    "type": "string",
                    "pattern": "^search:[0-9a-f]{24}$"
                  },
                  "kind": {
                    "const": "search"
                  },
                  "status": {
                    "enum": [
                      "pending",
                      "split",
                      "blocked_single_day_window",
                      "observed_pages_exhausted",
                      "partial_results",
                      "partial_api_window"
                    ]
                  },
                  "topic": {
                    "enum": [
                      "android-automation",
                      "browser-agent",
                      "browser-agents",
                      "browser-automation",
                      "computer-use",
                      "desktop-automation",
                      "gui-agent",
                      "gui-agents",
                      "gui-automation",
                      "computer-use-agent",
                      "computer-use-agents",
                      "gui-grounding",
                      "browser-use",
                      "web-agent",
                      "web-agents",
                      "desktop-agent",
                      "mobile-agent"
                    ]
                  },
                  "created_after": {
                    "type": "string",
                    "format": "date"
                  },
                  "created_before": {
                    "type": "string",
                    "format": "date"
                  },
                  "reported_total": {
                    "anyOf": [
                      {
                        "type": "integer",
                        "minimum": 0
                      },
                      {
                        "type": "null"
                      }
                    ]
                  },
                  "parent_id": {
                    "anyOf": [
                      {
                        "type": "string",
                        "pattern": "^(search:[0-9a-f]{24}|existing:[0-9a-f]{64})$"
                      },
                      {
                        "type": "null"
                      }
                    ]
                  }
                },
                "additionalProperties": false
              },
              {
                "type": "object",
                "required": [
                  "id",
                  "kind",
                  "status",
                  "input_id"
                ],
                "properties": {
                  "id": {
                    "type": "string",
                    "pattern": "^existing:[0-9a-f]{64}$"
                  },
                  "kind": {
                    "const": "existing"
                  },
                  "status": {
                    "enum": [
                      "pending",
                      "exhausted",
                      "superseded_input"
                    ]
                  },
                  "input_id": {
                    "type": "string",
                    "pattern": "^[0-9a-f]{64}$"
                  }
                },
                "additionalProperties": false
              }
            ]
          }
        },
        "request_outcomes": {
          "type": "object",
          "required": [],
          "properties": {
            "accepted": {
              "type": "integer",
              "minimum": 0
            },
            "accepted_with_rejections": {
              "type": "integer",
              "minimum": 0
            },
            "rejected": {
              "type": "integer",
              "minimum": 0
            },
            "rate_limited": {
              "type": "integer",
              "minimum": 0
            },
            "unavailable": {
              "type": "integer",
              "minimum": 0
            },
            "redirect_not_followed": {
              "type": "integer",
              "minimum": 0
            },
            "fetch_failed": {
              "type": "integer",
              "minimum": 0
            },
            "response_too_large": {
              "type": "integer",
              "minimum": 0
            },
            "malformed_response": {
              "type": "integer",
              "minimum": 0
            },
            "quarantined": {
              "type": "integer",
              "minimum": 0
            }
          },
          "additionalProperties": false
        },
        "global_completeness_established": {
          "const": false
        },
        "limit": {
          "const": "Bounded observed requests, not an exhaustive GitHub/CUA census. Per-query totals and overlapping shards are not a global union."
        },
        "epochs": {
          "type": "array",
          "minItems": 1,
          "items": {
            "$ref": "#/$defs/epoch"
          }
        },
        "current_epoch_id": {
          "type": "string",
          "pattern": "^epoch:[0-9a-f]{24}$"
        },
        "query_rechecks_count": {
          "type": "integer",
          "minimum": 0
        }
      },
      "additionalProperties": false,
      "dependentRequired": {
        "epochs": [
          "current_epoch_id",
          "query_rechecks_count"
        ],
        "current_epoch_id": [
          "epochs",
          "query_rechecks_count"
        ],
        "query_rechecks_count": [
          "epochs",
          "current_epoch_id"
        ]
      }
    },
    "active_catalog_context": {
      "type": "object",
      "required": [
        "sha256",
        "records",
        "role"
      ],
      "properties": {
        "sha256": {
          "type": "string",
          "pattern": "^[0-9a-f]{64}$"
        },
        "records": {
          "type": "integer",
          "minimum": 0
        },
        "role": {
          "const": "Comparison context only; not the observation source, selection, or approval."
        }
      },
      "additionalProperties": false
    },
    "count": {
      "type": "object",
      "required": [
        "value",
        "unit"
      ],
      "properties": {
        "value": {
          "type": "integer",
          "minimum": 0
        },
        "unit": {
          "const": "repositories"
        }
      },
      "additionalProperties": false
    },
    "supporting_primary_bodies": {
      "type": "integer",
      "minimum": 0
    },
    "observed_from": {
      "anyOf": [
        {
          "type": "string",
          "format": "date-time"
        },
        {
          "type": "null"
        }
      ]
    },
    "observed_through": {
      "anyOf": [
        {
          "type": "string",
          "format": "date-time"
        },
        {
          "type": "null"
        }
      ]
    },
    "review_policy": {
      "const": "Public GitHub metadata observations only. Every row remains candidate/relevance_pending; HTTP success, a topic tag, or a repository name does not establish CUA relevance, release completeness, licensing, or editorial approval."
    },
    "count_policy": {
      "const": "Unique positive numeric GitHub repository IDs within this observation snapshot. Query occurrences overlap. Candidates may overlap active catalog resources; these populations must not be added or treated as independent projects."
    },
    "privacy": {
      "type": "object",
      "required": [
        "checked_records",
        "omitted_fields",
        "policy"
      ],
      "properties": {
        "checked_records": {
          "type": "integer",
          "minimum": 0
        },
        "omitted_fields": {
          "type": "array",
          "items": {
            "type": "object",
            "required": [
              "id",
              "field",
              "reason"
            ],
            "properties": {
              "id": {
                "type": "string",
                "pattern": "^github-repository:[1-9][0-9]*$"
              },
              "field": {
                "const": "source_description"
              },
              "reason": {
                "enum": [
                  "credential_shaped_string",
                  "private_home_path",
                  "url_userinfo"
                ]
              }
            },
            "additionalProperties": false
          },
          "uniqueItems": true
        },
        "policy": {
          "const": "Only allowlisted public repository fields and accepted primary receipts. Suspect descriptions are explicitly omitted; owner/contact objects, headers, raw bodies and local paths are not exported."
        }
      },
      "additionalProperties": false
    },
    "export": {
      "type": "object",
      "required": [
        "file",
        "format",
        "bytes",
        "sha256"
      ],
      "properties": {
        "file": {
          "const": "github-candidates.jsonl"
        },
        "format": {
          "const": "JSONL"
        },
        "bytes": {
          "type": "integer",
          "minimum": 0
        },
        "sha256": {
          "type": "string",
          "pattern": "^[0-9a-f]{64}$"
        }
      },
      "additionalProperties": false
    },
    "schema": {
      "const": "/schemas/github-candidate-observation.schema.json"
    },
    "manifest_schema": {
      "const": "/schemas/github-candidate-observation-manifest.schema.json"
    },
    "verification": {
      "type": "object",
      "required": [
        "mode",
        "historical_raw_replayed",
        "baseline_observation_sha256",
        "baseline_checkpoint_sha256",
        "successful_receipts_replayed",
        "primary_bodies_replayed"
      ],
      "properties": {
        "mode": {
          "enum": [
            "full_retained_replay",
            "incremental_retained_replay"
          ]
        },
        "historical_raw_replayed": {
          "type": "boolean"
        },
        "baseline_observation_sha256": {
          "anyOf": [
            {
              "type": "string",
              "pattern": "^[0-9a-f]{64}$"
            },
            {
              "type": "null"
            }
          ]
        },
        "baseline_checkpoint_sha256": {
          "anyOf": [
            {
              "type": "string",
              "pattern": "^[0-9a-f]{64}$"
            },
            {
              "type": "null"
            }
          ]
        },
        "successful_receipts_replayed": {
          "type": "integer",
          "minimum": 0
        },
        "primary_bodies_replayed": {
          "type": "integer",
          "minimum": 0
        }
      },
      "additionalProperties": false
    },
    "checkpoint_state_sha256": {
      "anyOf": [
        {
          "type": "string",
          "pattern": "^[0-9a-f]{64}$"
        },
        {
          "type": "null"
        }
      ]
    }
  },
  "additionalProperties": false,
  "dependentRequired": {
    "verification": [
      "checkpoint_state_sha256"
    ],
    "checkpoint_state_sha256": [
      "verification"
    ]
  },
  "$defs": {
    "epoch": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "id",
        "configuration",
        "root_query_ids",
        "created_at",
        "reason",
        "previous_epoch_id",
        "source_state_sha256"
      ],
      "properties": {
        "id": {
          "type": "string",
          "pattern": "^epoch:[0-9a-f]{24}$"
        },
        "configuration": {
          "type": "object",
          "additionalProperties": false,
          "required": [
            "topics",
            "created_after",
            "created_before"
          ],
          "properties": {
            "topics": {
              "type": "array",
              "minItems": 1,
              "uniqueItems": true,
              "items": {
                "enum": [
                  "computer-use",
                  "gui-agent",
                  "gui-agents",
                  "gui-automation",
                  "desktop-automation",
                  "browser-agent",
                  "browser-agents",
                  "browser-automation",
                  "android-automation",
                  "computer-use-agent",
                  "computer-use-agents",
                  "gui-grounding",
                  "browser-use",
                  "web-agent",
                  "web-agents",
                  "desktop-agent",
                  "mobile-agent"
                ]
              }
            },
            "created_after": {
              "type": "string",
              "format": "date"
            },
            "created_before": {
              "type": "string",
              "format": "date"
            }
          }
        },
        "root_query_ids": {
          "type": "array",
          "minItems": 1,
          "uniqueItems": true,
          "items": {
            "type": "string",
            "pattern": "^search:[0-9a-f]{24}$"
          }
        },
        "created_at": {
          "type": "string",
          "format": "date-time"
        },
        "reason": {
          "enum": [
            "initial",
            "v1_migration",
            "topic_expansion",
            "daily_continuation",
            "frontier_continuation"
          ]
        },
        "previous_epoch_id": {
          "anyOf": [
            {
              "type": "string",
              "pattern": "^epoch:[0-9a-f]{24}$"
            },
            {
              "type": "null"
            }
          ]
        },
        "source_state_sha256": {
          "anyOf": [
            {
              "type": "string",
              "pattern": "^[0-9a-f]{64}$"
            },
            {
              "type": "null"
            }
          ]
        }
      }
    }
  }
}