Manifest

Every definition and field of the schema 3 Feed Package manifest (fruitful-package.json).

The Feed Package manifest (fruitful-package.json, schema 3). Four blocks in data-flow order: capture, transform, render, focus. Every schema the manifest refers to is a Lexicon; the manifest itself is version pins and wiring. Executable code is declared in exactly two places: capture.hooks and transform.hooks. Cross-field rules (a route's login exists, hook exports match the module, a binding's emits matches transform.input, a from route names a pinned record) are declared as data in FEED_PACKAGE_MANIFEST_RELATIONSHIPS and enforced by the registry, not by this document.

Add a $schema line to fruitful-package.json for validation and autocomplete. Fruitful ignores the key.

{ "$schema": "https://fruitful.app/schemas/fruitful-package.schema.json", "schema": 3, "kind": "page-plugin" }

Read top to bottom, the four blocks are the lifecycle of one followed page:

capture: a feed is offered (Explore, schedule), its route captures the page (login, load, ready, hooks, scroll), and the binding, through its extractor engine, produces the extract.
transform: the transform module turns the extract into Lexicon-validated records.
render: the root record becomes an Activity item through the package View and A2UI surfaces.
focus: independently of the pipeline, the browser extension hides or replaces distractions on the live site.

Executable code is declared in exactly two places: capture.hooks.file and transform.hooks.file. Rules that span fields are on the Relationships page; the engines a binding can name are on Extractor engines.

Examples come from the 33 committed first-party packages under feed-packages/.

main

A Feed Package manifest. Read top to bottom it is the lifecycle: a route is captured, the extract is transformed into records, records are rendered, and Focus is what the extension does on the live site meanwhile.

Field	Type	Required	Description
`schema`	`3`	yes	Manifest schema version. Schema 3 is the four-block layout; there is no dual-shape parsing.
`kind`	`"page-plugin"`	yes	Package kind discriminator shared with lexicon and extractor package manifests.
`name`	string	yes	Reverse-DNS package name, for example com.linkedin.posts. Immutable across versions.
`version`	string	yes	Semantic version of this immutable release, for example 2.0.0.
`runtime`	`"fruitful-page-plugin@1"`	yes	Module contract id for transform.hooks. It fixes the transform entry points (normalize, materializeRecords, buildView), so the manifest never restates them.
`files`	string[]	yes	Every file shipped in the package, as package-relative paths. Input to the release digest; a path referenced anywhere in the manifest must appear here.
`capture`	capture	yes	What can be followed, how each page is captured, and how the extract is produced.
`transform`	transform	yes	The code that turns an extract into records, and the record Lexicons it may emit.
`render`	render	no	How the root record becomes an Activity item. Required for follow and save feeds; absent for enrichment-only packages.
`focus`	focusPolicy[]	no	Browser-extension policies applied on the live site when a listed feed is followed. Independent of the capture pipeline; never runs package code.

Example:

{
  "schema": 3,
  "kind": "page-plugin",
  "name": "com.linkedin.posts",
  "version": "1.8.0",
  "runtime": "fruitful-page-plugin@1",
  "files": [
    "linkedin-page-plugin.js",
    "linkedin-capture-hooks.js",
    "presentation/compact.surface.json",
    "presentation/reader.surface.json"
  ],
  "capture": {
    "routes": [
      {
        "id": "newsfeed",
        "kind": "collection",
        "feed": {
          "name": "LinkedIn Newsfeed",
          "schedule": {
            "cron": "0 0 */4 * * *",
            "label": "Every 4 hours",
            "runOn": "desktop",
            "captureOnFollow": true
          },
          "explore": {
            "group": "feeds",
            "platform": "LinkedIn",
            "targets": [
              {
                "id": "linkedin-newsfeed",
                "title": "LinkedIn Newsfeed",
                "url": "https://www.linkedin.com/feed/"
              }
            ]
          }
        },
        "urls": [
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/feed"
            }
          },
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/feed/*"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/feed"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/feed/*"
            }
          }
        ],
        "runsOn": [
          "desktop"
        ],
        "login": "linkedin",
        "ready": {
          "selector": "[data-testid=\"mainFeed\"] div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
          "timeoutMs": 20000,
          "count": 1
        },
        "scroll": {
          "kind": "infinite",
          "container": "[data-testid=\"mainFeed\"]",
          "item": "div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
          "minItems": 60,
          "maxAttempts": 25
        },
        "binding": "newsfeed"
      },
      {
        "id": "company-posts",
        "kind": "collection",
        "feed": {
          "name": "LinkedIn Company Posts",
          "schedule": {
            "cron": "0 0 9 * * *",
            "label": "Daily",
            "runOn": "desktop",
            "captureOnFollow": true
          },
          "explore": {
            "group": "people-companies",
            "platform": "LinkedIn",
            "targets": [
              {
                "id": "elevenlabs-linkedin-posts",
                "title": "ElevenLabs LinkedIn Posts",
                "url": "https://www.linkedin.com/company/elevenlabs/posts/",
                "input": {
                  "label": "Company slug",
                  "placeholder": "elevenlabs",
                  "urlTemplate": "https://www.linkedin.com/company/{value}/posts/",
                  "valuePattern": "^[A-Za-z0-9][A-Za-z0-9-]{0,99}$"
                }
              }
            ]
          }
        },
        "urls": [
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/company/:slug/posts"
            }
          },
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/company/:slug/posts/"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/company/:slug/posts"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/company/:slug/posts/"
            }
          }
        ],
        "runsOn": [
          "desktop"
        ],
        "login": "linkedin",
        "ready": {
          "selector": ".feed-shared-update-v2",
          "timeoutMs": 20000,
          "count": 1
        },
        "on": {
          "afterReady": {
            "hook": "prepareCompanyPosts",
            "onError": "continue"
          }
        },
        "scroll": {
          "kind": "finite",
          "maxAttempts": 5
        },
        "binding": "classic"
      },
      {
        "id": "profile-posts",
        "kind": "collection",
        "feed": {
          "name": "LinkedIn Profile Posts",
          "schedule": {
            "cron": "0 0 9 * * *",
            "label": "Daily",
            "runOn": "desktop",
            "captureOnFollow": true
          },
          "explore": {
            "group": "people-companies",
            "platform": "LinkedIn",
            "targets": [
              {
                "id": "satya-nadella-linkedin-posts",
                "title": "Satya Nadella LinkedIn Posts",
                "url": "https://www.linkedin.com/in/satyanadella/recent-activity/all/",
                "input": {
                  "label": "Profile slug",
                  "placeholder": "satyanadella",
                  "urlTemplate": "https://www.linkedin.com/in/{value}/recent-activity/all/",
                  "valuePattern": "^[A-Za-z0-9][A-Za-z0-9-]{0,99}$"
                }
              }
            ]
          }
        },
        "urls": [
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/in/:slug/recent-activity"
            }
          },
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/in/:slug/recent-activity/*"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/in/:slug/recent-activity"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/in/:slug/recent-activity/*"
            }
          }
        ],
        "runsOn": [
          "desktop"
        ],
        "login": "linkedin",
        "ready": {
          "selector": ".feed-shared-update-v2",
          "timeoutMs": 20000,
          "count": 1
        },
        "scroll": {
          "kind": "finite",
          "maxAttempts": 5
        },
        "binding": "classic"
      }
    ],
    "logins": [
      {
        "id": "linkedin",
        "name": "LinkedIn",
        "reason": "LinkedIn captures need a logged-in browser session.",
        "checkUrl": "https://www.linkedin.com/in/me/",
        "signInUrl": "https://www.linkedin.com/login",
        "signedIn": {
          "selectors": [
            "[data-testid=\"mainFeed\"]",
            "[componentkey^=\"com.linkedin.sdui.profile.card.ref\"]",
            "section[data-view-name=\"profile-card\"]"
          ]
        },
        "signedOut": {
          "reason": "LinkedIn redirected to a sign-in or checkpoint page.",
          "urls": [
            {
              "pattern": {
                "hostname": "linkedin.com",
                "pathname": "/uas/login*"
              }
            },
            {
              "pattern": {
                "hostname": "linkedin.com",
                "pathname": "/checkpoint/*"
              }
            },
            {
              "pattern": {
                "hostname": "linkedin.com",
                "pathname": "/login*"
              }
            },
            {
              "pattern": {
                "hostname": "*.linkedin.com",
                "pathname": "/uas/login*"
              }
            },
            {
              "pattern": {
                "hostname": "*.linkedin.com",
                "pathname": "/checkpoint/*"
              }
            },
            {
              "pattern": {
                "hostname": "*.linkedin.com",
                "pathname": "/login*"
              }
            }
          ],
          "html": [
            "id=\"rememberme-div\"",
            "action=\"/checkpoint/rm/login-submit\""
          ]
        }
      }
    ],
    "hooks": {
      "file": "linkedin-capture-hooks.js",
      "exports": [
        {
          "name": "prepareCompanyPosts",
          "capabilities": [
            "browser.dom.read",
            "browser.dom.click"
          ]
        }
      ]
    },
    "bindings": [
      {
        "id": "classic",
        "package": "com.linkedin.posts.binding.classic@^2.1.0"
      },
      {
        "id": "newsfeed",
        "package": "com.linkedin.posts.binding.newsfeed@^2.1.0"
      }
    ]
  },
  "transform": {
    "hooks": {
      "file": "linkedin-page-plugin.js"
    },
    "records": [
      "app.fruitful.identity.linkedinAccount@^1.0.0",
      "app.fruitful.content.linkedinPost@^1.0.0",
      "app.fruitful.activity.repost@^1.0.0"
    ]
  },
  "render": {
    "root": "app.fruitful.content.linkedinPost",
    "view": "app.fruitful.feed.linkedin.defs@^1.0.0#postView",
    "occurredAt": "/publishedAt",
    "headline": {
      "path": "/text",
      "default": "LinkedIn post"
    },
    "surfaces": {
      "compact": "presentation/compact.surface.json",
      "expanded": "presentation/reader.surface.json"
    },
    "actions": [
      {
        "key": "linkedin-browser",
        "label": "LinkedIn",
        "view": "fruitful.browser.v1",
        "url": "/record/url",
        "rel": [
          "canonical"
        ]
      }
    ]
  },
  "focus": [
    {
      "id": "newsfeed-focus",
      "feeds": [
        "newsfeed"
      ],
      "urls": [
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/feed"
          }
        },
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/feed/*"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/feed"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/feed/*"
          }
        }
      ],
      "hide": [
        "www.linkedin.com##aside.scaffold-layout__aside",
        "linkedin.com##aside.scaffold-layout__aside",
        "www.linkedin.com##.feed-right-rail",
        "linkedin.com##.feed-right-rail",
        "www.linkedin.com##[data-test-id=\"feed-right-rail\"]",
        "linkedin.com##[data-test-id=\"feed-right-rail\"]",
        "www.linkedin.com##section:has(h2:has-text(/^LinkedIn News$/i))",
        "linkedin.com##section:has(h2:has-text(/^LinkedIn News$/i))"
      ],
      "replace": {
        "selector": "[data-testid=\"mainFeed\"]",
        "with": "panel",
        "title": "Read your LinkedIn updates in Fruitful"
      }
    }
  ]
}

capture

The capture block. Routes are the unit; a route's feed is how it is offered to people, a route's detailOf is which master it details; hooks are the only capture-time code; bindings configure the platform extractor engines that turn a page into records.

Field	Type	Required	Description
`routes`	route[]	yes	Every page this package knows how to capture, one route each. A route says what it yields (kind), how it is reached (feed for people, detailOf for entries of a record type, neither for a pasted URL), and how the page becomes a record (binding). URL resolution across routes must be unambiguous.
`logins`	login[]	no	Signed-in browser sessions a route may require. Referenced from routes by id.
`hooks`	captureHooks	no	The capture-hook module and the capabilities each export is granted. One of the two places code may appear.
`bindings`	bindingDependency[]	yes	Binding packages this package pins, each with the alias routes use. A binding configures one platform extractor engine (dom, reader, media, rss, http) for one site; first-party bindings such as app.fruitful.binding.reader are pinned like any other.

Example:

{
  "routes": [
    {
      "id": "newsfeed",
      "kind": "collection",
      "feed": {
        "name": "LinkedIn Newsfeed",
        "schedule": {
          "cron": "0 0 */4 * * *",
          "label": "Every 4 hours",
          "runOn": "desktop",
          "captureOnFollow": true
        },
        "explore": {
          "group": "feeds",
          "platform": "LinkedIn",
          "targets": [
            {
              "id": "linkedin-newsfeed",
              "title": "LinkedIn Newsfeed",
              "url": "https://www.linkedin.com/feed/"
            }
          ]
        }
      },
      "urls": [
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/feed"
          }
        },
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/feed/*"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/feed"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/feed/*"
          }
        }
      ],
      "runsOn": [
        "desktop"
      ],
      "login": "linkedin",
      "ready": {
        "selector": "[data-testid=\"mainFeed\"] div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
        "timeoutMs": 20000,
        "count": 1
      },
      "scroll": {
        "kind": "infinite",
        "container": "[data-testid=\"mainFeed\"]",
        "item": "div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
        "minItems": 60,
        "maxAttempts": 25
      },
      "binding": "newsfeed"
    },
    {
      "id": "company-posts",
      "kind": "collection",
      "feed": {
        "name": "LinkedIn Company Posts",
        "schedule": {
          "cron": "0 0 9 * * *",
          "label": "Daily",
          "runOn": "desktop",
          "captureOnFollow": true
        },
        "explore": {
          "group": "people-companies",
          "platform": "LinkedIn",
          "targets": [
            {
              "id": "elevenlabs-linkedin-posts",
              "title": "ElevenLabs LinkedIn Posts",
              "url": "https://www.linkedin.com/company/elevenlabs/posts/",
              "input": {
                "label": "Company slug",
                "placeholder": "elevenlabs",
                "urlTemplate": "https://www.linkedin.com/company/{value}/posts/",
                "valuePattern": "^[A-Za-z0-9][A-Za-z0-9-]{0,99}$"
              }
            }
          ]
        }
      },
      "urls": [
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/company/:slug/posts"
          }
        },
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/company/:slug/posts/"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/company/:slug/posts"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/company/:slug/posts/"
          }
        }
      ],
      "runsOn": [
        "desktop"
      ],
      "login": "linkedin",
      "ready": {
        "selector": ".feed-shared-update-v2",
        "timeoutMs": 20000,
        "count": 1
      },
      "on": {
        "afterReady": {
          "hook": "prepareCompanyPosts",
          "onError": "continue"
        }
      },
      "scroll": {
        "kind": "finite",
        "maxAttempts": 5
      },
      "binding": "classic"
    },
    {
      "id": "profile-posts",
      "kind": "collection",
      "feed": {
        "name": "LinkedIn Profile Posts",
        "schedule": {
          "cron": "0 0 9 * * *",
          "label": "Daily",
          "runOn": "desktop",
          "captureOnFollow": true
        },
        "explore": {
          "group": "people-companies",
          "platform": "LinkedIn",
          "targets": [
            {
              "id": "satya-nadella-linkedin-posts",
              "title": "Satya Nadella LinkedIn Posts",
              "url": "https://www.linkedin.com/in/satyanadella/recent-activity/all/",
              "input": {
                "label": "Profile slug",
                "placeholder": "satyanadella",
                "urlTemplate": "https://www.linkedin.com/in/{value}/recent-activity/all/",
                "valuePattern": "^[A-Za-z0-9][A-Za-z0-9-]{0,99}$"
              }
            }
          ]
        }
      },
      "urls": [
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/in/:slug/recent-activity"
          }
        },
        {
          "pattern": {
            "hostname": "linkedin.com",
            "pathname": "/in/:slug/recent-activity/*"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/in/:slug/recent-activity"
          }
        },
        {
          "pattern": {
            "hostname": "*.linkedin.com",
            "pathname": "/in/:slug/recent-activity/*"
          }
        }
      ],
      "runsOn": [
        "desktop"
      ],
      "login": "linkedin",
      "ready": {
        "selector": ".feed-shared-update-v2",
        "timeoutMs": 20000,
        "count": 1
      },
      "scroll": {
        "kind": "finite",
        "maxAttempts": 5
      },
      "binding": "classic"
    }
  ],
  "logins": [
    {
      "id": "linkedin",
      "name": "LinkedIn",
      "reason": "LinkedIn captures need a logged-in browser session.",
      "checkUrl": "https://www.linkedin.com/in/me/",
      "signInUrl": "https://www.linkedin.com/login",
      "signedIn": {
        "selectors": [
          "[data-testid=\"mainFeed\"]",
          "[componentkey^=\"com.linkedin.sdui.profile.card.ref\"]",
          "section[data-view-name=\"profile-card\"]"
        ]
      },
      "signedOut": {
        "reason": "LinkedIn redirected to a sign-in or checkpoint page.",
        "urls": [
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/uas/login*"
            }
          },
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/checkpoint/*"
            }
          },
          {
            "pattern": {
              "hostname": "linkedin.com",
              "pathname": "/login*"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/uas/login*"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/checkpoint/*"
            }
          },
          {
            "pattern": {
              "hostname": "*.linkedin.com",
              "pathname": "/login*"
            }
          }
        ],
        "html": [
          "id=\"rememberme-div\"",
          "action=\"/checkpoint/rm/login-submit\""
        ]
      }
    }
  ],
  "hooks": {
    "file": "linkedin-capture-hooks.js",
    "exports": [
      {
        "name": "prepareCompanyPosts",
        "capabilities": [
          "browser.dom.read",
          "browser.dom.click"
        ]
      }
    ]
  },
  "bindings": [
    {
      "id": "classic",
      "package": "com.linkedin.posts.binding.classic@^2.1.0"
    },
    {
      "id": "newsfeed",
      "package": "com.linkedin.posts.binding.newsfeed@^2.1.0"
    }
  ]
}

feed

How a route is offered to people. A collection with a feed is followed on its schedule; an item with a feed is saved once. Routes without a feed never appear in Explore or Activity.

Field	Type	Required	Description
`name`	string	yes	Display name shown in intake, Registry, and Explore.
`schedule`	schedule	no	Capture cadence and placement applied when a person follows. Only meaningful on collection routes.
`explore`	explore	no	How the feed appears in Explore. Omit to keep the feed out of the catalog.

Example:

{
  "name": "LinkedIn Newsfeed",
  "schedule": {
    "cron": "0 0 */4 * * *",
    "label": "Every 4 hours",
    "runOn": "desktop",
    "captureOnFollow": true
  },
  "explore": {
    "group": "feeds",
    "platform": "LinkedIn",
    "targets": [
      {
        "id": "linkedin-newsfeed",
        "title": "LinkedIn Newsfeed",
        "url": "https://www.linkedin.com/feed/"
      }
    ]
  }
}

schedule

Defaults applied to the page created when a user follows a feed.

Field	Type	Required	Description
`cron`	string	yes	Six-field cron expression for recurring captures.
`label`	string	no	Human label for the cadence shown in Explore and intake, for example Every 4 hours.
`runOn`	`"desktop"` \| `"cloud"`	no	Where captures run. desktop uses the user's signed-in browser via the Desktop app; cloud uses the screenshot service.
`captureOnFollow`	boolean	no	Whether to capture immediately when the feed is followed.

Example:

{
  "cron": "0 0 */4 * * *",
  "label": "Every 4 hours",
  "runOn": "desktop",
  "captureOnFollow": true
}

explore

Explore catalog placement for a feed.

Field	Type	Required	Description
`group`	`"feeds"` \| `"people-companies"` \| `"communities"` \| `"media"`	yes	Explore section the targets are listed under.
`platform`	string	yes	Platform label on the Explore card, for example LinkedIn.
`targets`	exploreTarget[]	yes	Concrete URLs a user can pick, optionally parameterised.

Example:

{
  "group": "feeds",
  "platform": "LinkedIn",
  "targets": [
    {
      "id": "linkedin-newsfeed",
      "title": "LinkedIn Newsfeed",
      "url": "https://www.linkedin.com/feed/"
    }
  ]
}

exploreTarget

One Explore card.

Field	Type	Required	Description
`id`	string	yes	Stable target id, kebab-case.
`title`	string	yes	Card title.
`url`	string (uri)	yes	Example or default URL. Must match the feed's route patterns.
`input`	exploreInput	no	Present when the user supplies a value (a handle, a slug) that is substituted into a URL template.

Example:

{
  "id": "linkedin-newsfeed",
  "title": "LinkedIn Newsfeed",
  "url": "https://www.linkedin.com/feed/"
}

exploreInput

A single user-supplied value that builds the target URL.

Field	Type	Required	Description
`label`	string	yes	Field label, for example Company slug.
`placeholder`	string	yes	Example value shown in the field.
`urlTemplate`	string	yes	URL with a {value} placeholder.
`valuePattern`	string	no	Regular expression the raw value must match before substitution.
`prefixToStrip`	string	no	Prefix removed from the raw value before substitution, for example @.

Example:

{
  "label": "Company slug",
  "placeholder": "elevenlabs",
  "urlTemplate": "https://www.linkedin.com/company/{value}/posts/",
  "valuePattern": "^[A-Za-z0-9][A-Za-z0-9-]{0,99}$"
}

A signed-in browser session and how to recognise it.

Field	Type	Required	Description
`id`	string	yes	Login id referenced by routes.
`name`	string	yes	Display name, for example LinkedIn.
`reason`	string	no	Shown to the user when explaining why sign-in is needed.
`checkUrl`	string (uri)	yes	URL loaded to test the session.
`signInUrl`	string (uri)	yes	URL opened for the user to sign in.
`signedIn`	sessionMatcher	no	Positive match: the session is ready when this matches the check page.
`signedOut`	sessionMatcher	no	Negative match: the session is missing when this matches the final URL or HTML.

Example:

{
  "id": "linkedin",
  "name": "LinkedIn",
  "reason": "LinkedIn captures need a logged-in browser session.",
  "checkUrl": "https://www.linkedin.com/in/me/",
  "signInUrl": "https://www.linkedin.com/login",
  "signedIn": {
    "selectors": [
      "[data-testid=\"mainFeed\"]",
      "[componentkey^=\"com.linkedin.sdui.profile.card.ref\"]",
      "section[data-view-name=\"profile-card\"]"
    ]
  },
  "signedOut": {
    "reason": "LinkedIn redirected to a sign-in or checkpoint page.",
    "urls": [
      {
        "pattern": {
          "hostname": "linkedin.com",
          "pathname": "/uas/login*"
        }
      },
      {
        "pattern": {
          "hostname": "linkedin.com",
          "pathname": "/checkpoint/*"
        }
      },
      {
        "pattern": {
          "hostname": "linkedin.com",
          "pathname": "/login*"
        }
      },
      {
        "pattern": {
          "hostname": "*.linkedin.com",
          "pathname": "/uas/login*"
        }
      },
      {
        "pattern": {
          "hostname": "*.linkedin.com",
          "pathname": "/checkpoint/*"
        }
      },
      {
        "pattern": {
          "hostname": "*.linkedin.com",
          "pathname": "/login*"
        }
      }
    ],
    "html": [
      "id=\"rememberme-div\"",
      "action=\"/checkpoint/rm/login-submit\""
    ]
  }
}

sessionMatcher

Any of the listed conditions matching counts as a match.

Field	Type	Required	Description
`reason`	string	no	Message shown when this matcher fires.
`urls`	urlPattern[]	no	URL patterns tested against the final URL.
`selectors`	string[]	no	CSS selectors tested against the page.
`html`	string[]	no	Literal substrings tested against the page HTML.

Example:

{
  "selectors": [
    "[data-testid=\"mainFeed\"]",
    "[componentkey^=\"com.linkedin.sdui.profile.card.ref\"]",
    "section[data-view-name=\"profile-card\"]"
  ]
}

captureHooks

The capture-hook module. Every export is a function the platform calls at a named lifecycle point with exactly the capabilities declared here; the declaration becomes the sandbox import table.

Field	Type	Required	Description
`file`	string	yes	Package-relative path to the self-contained hook bundle. Must appear in files.
`stdlib`	string (known: `url`, `intl`, `base64`)[]	no	Host-provided pure modules linked into the sandbox. Declared, then linked; never inferred from the bundle.
`exports`	hookExport[]	no	Capability grants per export. Every hook referenced by a route must be listed; an unlisted export gets an empty import table. Publish validation checks each name exists in the module.

Example:

{
  "file": "linkedin-capture-hooks.js",
  "exports": [
    {
      "name": "prepareCompanyPosts",
      "capabilities": [
        "browser.dom.read",
        "browser.dom.click"
      ]
    }
  ]
}

hookExport

One exported hook and the authority it is granted.

Field	Type	Required	Description
`name`	string	yes	Export name in the hook module.
`capabilities`	(`"browser.dom.read"` \| `"browser.dom.click"` \| `"browser.dom.type"`)[]	yes	Capability ids; each maps to a fixed set of page operations in the SafePage import table.

Example:

{
  "name": "prepareCompanyPosts",
  "capabilities": [
    "browser.dom.read",
    "browser.dom.click"
  ]
}

route

One page the package can capture. kind says what the route yields: a collection (the master: many entries that get ids and become Activity, re-observed on a schedule) or an item (the detail: one thing that stands alone and enriches an entry). binding says how the page becomes a record: which pinned binding, and through it which platform extractor engine, runs over it.

Field	Type	Required	Description
`id`	string	yes	Stable route identity, kebab-case. Pinned into installs and provenance when the route carries a feed; never renamed.
`kind`	`"collection"` \| `"item"`	yes	collection: the page lists many entries; each gets a stable id and new ones become Activity. item: the page is one thing (an article, a post, a video) that stands alone and enriches an entry.
`feed`	feed	no	Offer this route to people. Present on routes that appear in Explore and can be followed or saved.
`detailOf`	recordSource	no	The master this item is the detail of: each entry of the named record type has a URL at the given pointer, and that URL is captured through this route, one hop, bounded by the platform.
`urls`	urlPattern[]	no	URL patterns this route handles. Required unless the route is the detail of a record type, where it may constrain which entry URLs are accepted.
`runsOn`	(`"desktop"` \| `"cloud"`)[]	yes	Where this route may execute. Authoring tooling can run any route and is not listed.
`login`	string	no	Id of a login that must be ready before capture.
`load`	loadOptions	no	Navigation wait and timeout. Defaults: domcontentloaded, 45000 ms.
`ready`	readiness	no	Selector the page must satisfy before hooks and scrolling run.
`on`	lifecycle	no	Capture hooks attached to lifecycle points. Omission is the canonical empty.
`scroll`	scroll	no	Scroll strategy applied before the snapshot.
`binding`	string	yes	Alias from capture.bindings. The binding's extractor engine decides what the page becomes: a dom binding emits the transform's input, a reader binding emits a webDocument, a media binding emits a media record. Collections need a binding that emits transform.input; reader and media bindings are for items.

Example:

{
  "id": "newsfeed",
  "kind": "collection",
  "feed": {
    "name": "LinkedIn Newsfeed",
    "schedule": {
      "cron": "0 0 */4 * * *",
      "label": "Every 4 hours",
      "runOn": "desktop",
      "captureOnFollow": true
    },
    "explore": {
      "group": "feeds",
      "platform": "LinkedIn",
      "targets": [
        {
          "id": "linkedin-newsfeed",
          "title": "LinkedIn Newsfeed",
          "url": "https://www.linkedin.com/feed/"
        }
      ]
    }
  },
  "urls": [
    {
      "pattern": {
        "hostname": "linkedin.com",
        "pathname": "/feed"
      }
    },
    {
      "pattern": {
        "hostname": "linkedin.com",
        "pathname": "/feed/*"
      }
    },
    {
      "pattern": {
        "hostname": "*.linkedin.com",
        "pathname": "/feed"
      }
    },
    {
      "pattern": {
        "hostname": "*.linkedin.com",
        "pathname": "/feed/*"
      }
    }
  ],
  "runsOn": [
    "desktop"
  ],
  "login": "linkedin",
  "ready": {
    "selector": "[data-testid=\"mainFeed\"] div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
    "timeoutMs": 20000,
    "count": 1
  },
  "scroll": {
    "kind": "infinite",
    "container": "[data-testid=\"mainFeed\"]",
    "item": "div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
    "minItems": 60,
    "maxAttempts": 25
  },
  "binding": "newsfeed"
}

loadOptions

Navigation settings.

Field	Type	Required	Description
`waitUntil`	`"domcontentloaded"` \| `"load"`	no	Navigation event to wait for.
`timeoutMs`	integer	no	Navigation timeout in milliseconds.

readiness

The condition that marks the page as loaded enough to capture.

Field	Type	Required	Description
`selector`	string	yes	CSS selector to wait for.
`count`	integer	no	Minimum number of matches.
`timeoutMs`	integer	no	How long to wait before the route fails.

Example:

{
  "selector": "[data-testid=\"mainFeed\"] div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
  "timeoutMs": 20000,
  "count": 1
}

lifecycle

Hook attachments by capture lifecycle point. The point vocabulary is the platform's; unknown points are rejected.

Field	Type	Required	Description
`afterLoad`	hookRef	no	After navigation completes, before the readiness wait.
`afterReady`	hookRef	no	After the readiness selector is satisfied, before scrolling.
`beforeSnapshot`	hookRef	no	After scrolling settles, immediately before the snapshot.

Example:

{
  "afterReady": {
    "hook": "prepareCompanyPosts",
    "onError": "continue"
  }
}

hookRef

A reference to a capture hook export plus per-usage failure policy.

Field	Type	Required	Description
`hook`	string	yes	Export name declared in capture.hooks.exports.
`onError`	`"fail"` \| `"continue"`	no	fail aborts the capture; continue proceeds to the snapshot. Per usage because the same hook can be load-bearing on one route and best-effort on another.

Example:

{
  "hook": "prepareCompanyPosts",
  "onError": "continue"
}

scroll

Scroll strategy. kind selects the algorithm; container and item are required for infinite, stitching and paginate, next is required for paginate, and all are ignored for finite (enforced by the registry).

Field	Type	Required	Description
`kind`	`"finite"` \| `"infinite"` \| `"stitching"` \| `"paginate"`	yes	finite scrolls to the bottom a bounded number of times; infinite scrolls a container until item growth stops; stitching also concatenates virtualised items that leave the DOM; paginate follows the next link and stitches each page's items.
`container`	string	no	CSS selector of the scrolling container.
`item`	string	no	CSS selector of one feed item, used to measure growth.
`minItems`	integer	no	Stop once this many items are present.
`maxAttempts`	integer	no	Upper bound on scroll iterations.
`delayMs`	integer	no	Pause between iterations.
`itemGrowthTimeoutMs`	integer	no	Give up when no new items appear within this window.
`dedupeAttribute`	string	no	stitching only: attribute that identifies an item across virtualisation.
`next`	string	no	paginate only: CSS selector of the next-page link. Its href is followed until it is absent, minItems is reached, or maxPages is hit.
`maxPages`	integer	no	paginate only: upper bound on pages visited, the first page included. Defaults to 5.

Example:

{
  "kind": "infinite",
  "container": "[data-testid=\"mainFeed\"]",
  "item": "div[role=\"listitem\"]:has([componentkey*='-replaceableCommentTools'])",
  "minItems": 60,
  "maxAttempts": 25
}

urlPattern

A URLPattern plus sibling constraints. The pattern wrapper is deliberate: exclude, groups, and params are siblings, not children, of the pattern.

Field	Type	Required	Description
`pattern`	urlPatternInit	yes	URLPattern init; omitted components match anything.
`exclude`	urlPatternInit[]	no	Patterns that reject a URL even when pattern matches.
`groups`	pathnameGroup[]	no	Constraints on named pathname groups such as :handle.
`params`	searchParam[]	no	Constraints on query parameters.

Example:

{
  "pattern": {
    "hostname": "linkedin.com",
    "pathname": "/feed"
  }
}

urlPatternInit

URLPattern components, each in URLPattern syntax.

Field	Type	Required
`protocol`	string	no
`username`	string	no
`password`	string	no
`hostname`	string	no
`port`	string	no
`pathname`	string	no
`search`	string	no
`hash`	string	no
`baseURL`	string (uri)	no

Example:

{
  "hostname": "linkedin.com",
  "pathname": "/feed"
}

pathnameGroup

A constraint on one named pathname group.

Field	Type	Required	Description
`name`	string	yes	Group name without the leading colon.
`regex`	string	no	Regular expression the group value must match.
`oneOf`	string[]	no	Allowed values.
`caseInsensitive`	boolean	no	Compare ignoring case.

Example:

{
  "name": "handle",
  "regex": "^[a-zA-Z_][a-zA-Z0-9_]{0,20}$"
}

searchParam

A constraint on one query parameter.

Field	Type	Required	Description
`name`	string	yes	Parameter name.
`required`	boolean	no	Whether the parameter must be present.
`regex`	string	no	Regular expression the value must match.
`oneOf`	string[]	no	Allowed values.
`caseInsensitive`	boolean	no	Compare ignoring case.

transform

The transform block: the input def, the module that turns an extract into records, and the record Lexicons it may emit.

Field	Type	Required	Description
`input`	string	no	The Lexicon def the transform consumes, as nsid@range#def. Optional when every dom, rss or http binding the package pins emits the same def, in which case it is that def; required when they differ, and then the transform must export normalize for every binding whose emits is not this def.
`hooks`	transformHooks	yes	The transform module. The second of the two places code may appear.
`records`	string[]	yes	Record Lexicon pins as nsid@range. Identity is recordRef.uri and links are ref-typed properties, both read from the Lexicon; nothing is restated here.

Example:

{
  "hooks": {
    "file": "linkedin-page-plugin.js"
  },
  "records": [
    "app.fruitful.identity.linkedinAccount@^1.0.0",
    "app.fruitful.content.linkedinPost@^1.0.0",
    "app.fruitful.activity.repost@^1.0.0"
  ]
}

transformHooks

The transform module. Entry points are fixed by runtime: normalize (only when a binding emits a def other than transform.input), materializeRecords, buildView.

Field	Type	Required	Description
`file`	string	yes	Package-relative path to the self-contained transform bundle. Must appear in files and differ from capture.hooks.file.
`stdlib`	string (known: `url`, `intl`, `base64`)[]	no	Host-provided pure modules linked into the sandbox.

Example:

{
  "file": "linkedin-page-plugin.js"
}

render

The render block. Declares which record is the Activity root and how the trusted catalog presents it. Never contains code.

Field	Type	Required	Description
`root`	string (nsid)	yes	NSID of the record Lexicon that becomes an Activity item. Must appear in transform.records.
`view`	string	yes	View Lexicon def buildView must produce, as nsid@range#def.
`occurredAt`	string	no	JSON Pointer into the root record for the item's timestamp. Omit to use the capture time; when present and the path is absent from a record, capture time is the fallback.
`headline`	headline	yes	Where the Activity shell headline comes from.
`surfaces`	surfaces	yes	A2UI surface templates. Each carries its own catalog id; they must agree.
`actions`	action[]	no	Alternate ways to open the item, resolved against the validated View.

Example:

{
  "root": "app.fruitful.content.linkedinPost",
  "view": "app.fruitful.feed.linkedin.defs@^1.0.0#postView",
  "occurredAt": "/publishedAt",
  "headline": {
    "path": "/text",
    "default": "LinkedIn post"
  },
  "surfaces": {
    "compact": "presentation/compact.surface.json",
    "expanded": "presentation/reader.surface.json"
  },
  "actions": [
    {
      "key": "linkedin-browser",
      "label": "LinkedIn",
      "view": "fruitful.browser.v1",
      "url": "/record/url",
      "rel": [
        "canonical"
      ]
    }
  ]
}

headline

Headline source for the Activity shell. With path, the record value is used and default is the fallback; without path, default is the static headline.

Field	Type	Required	Description
`path`	string	no	JSON Pointer into the root record. Omit for a static headline.
`default`	string	yes	Fallback when path is absent or empty, or the static headline when there is no path.

Example:

{
  "path": "/text",
  "default": "LinkedIn post"
}

surfaces

Package-authored A2UI v0.9.1 surface templates.

Field	Type	Required	Description
`compact`	string	yes	Package-relative path to the compact surface JSON.
`expanded`	string	yes	Package-relative path to the expanded (reader) surface JSON.

Example:

{
  "compact": "presentation/compact.surface.json",
  "expanded": "presentation/reader.surface.json"
}

action

A resource the item can be opened as, for example the original website.

Field	Type	Required	Description
`key`	string	yes	Stable key, kebab-case.
`label`	string	yes	Button label.
`view`	string	yes	Trusted view type, for example fruitful.browser.v1 or fruitful.webReader.v1.
`url`	string	yes	JSON Pointer into the View selecting the resource URI.
`mediaType`	string	no	MIME type of the resource when known.
`rel`	string (known: `canonical`, `replies`)[]	no	Relationships from the item to the resource; known values or absolute extension URIs.

Example:

{
  "key": "linkedin-browser",
  "label": "LinkedIn",
  "view": "fruitful.browser.v1",
  "url": "/record/url",
  "rel": [
    "canonical"
  ]
}

focusPolicy

A browser-extension Focus policy. Active only while a listed feed is followed by an installed package.

Field	Type	Required	Description
`id`	string	yes	Policy id, kebab-case.
`feeds`	string[]	yes	Ids of routes whose feed being followed activates this policy.
`urls`	urlPattern[]	yes	Browsing URLs the policy applies to. Distinct from capture routes.
`hide`	string[]	no	Adblock-syntax cosmetic rules applied by the extension's cosmetic engine.
`replace`	replacement	no	Replace a distraction surface with a Fruitful panel or interstitial.

Example:

{
  "id": "newsfeed-focus",
  "feeds": [
    "newsfeed"
  ],
  "urls": [
    {
      "pattern": {
        "hostname": "linkedin.com",
        "pathname": "/feed"
      }
    },
    {
      "pattern": {
        "hostname": "linkedin.com",
        "pathname": "/feed/*"
      }
    },
    {
      "pattern": {
        "hostname": "*.linkedin.com",
        "pathname": "/feed"
      }
    },
    {
      "pattern": {
        "hostname": "*.linkedin.com",
        "pathname": "/feed/*"
      }
    }
  ],
  "hide": [
    "www.linkedin.com##aside.scaffold-layout__aside",
    "linkedin.com##aside.scaffold-layout__aside",
    "www.linkedin.com##.feed-right-rail",
    "linkedin.com##.feed-right-rail",
    "www.linkedin.com##[data-test-id=\"feed-right-rail\"]",
    "linkedin.com##[data-test-id=\"feed-right-rail\"]",
    "www.linkedin.com##section:has(h2:has-text(/^LinkedIn News$/i))",
    "linkedin.com##section:has(h2:has-text(/^LinkedIn News$/i))"
  ],
  "replace": {
    "selector": "[data-testid=\"mainFeed\"]",
    "with": "panel",
    "title": "Read your LinkedIn updates in Fruitful"
  }
}

replacement

Replaces the matched element in the live page.

Field	Type	Required	Description
`selector`	string	yes	CSS selector of the element to replace.
`with`	`"panel"` \| `"interstitial"`	yes	panel replaces a feed with a Fruitful panel; interstitial replaces a whole surface and pauses media inside it.
`title`	string	no	Panel title.
`description`	string	no	Panel body text.

Example:

{
  "selector": "[data-testid=\"mainFeed\"]",
  "with": "panel",
  "title": "Read your LinkedIn updates in Fruitful"
}

recordSource

Which master an item route is the detail of: the record type whose entries carry the URL, read off each root record after the transform runs.

Field	Type	Required	Description
`record`	string (nsid)	yes	NSID of the record Lexicon whose instances carry the URL. Must be the render root, since the frontier is read off Activity items.
`url`	string	yes	JSON Pointer into the record selecting a URI string, for example /destinationUrl.

Example:

{
  "record": "app.fruitful.feed.hackerNews.submission",
  "url": "/destinationUrl"
}

bindingDependency

A pinned binding package, addressed from routes by alias.

Field	Type	Required	Description
`id`	string	yes	Alias used in route.binding.
`package`	string	yes	Binding package pin as name@range, for example com.linkedin.posts.binding.newsfeed@^2.1.0 or app.fruitful.binding.reader@^1.0.0.

Example:

{
  "id": "classic",
  "package": "com.linkedin.posts.binding.classic@^2.1.0"
}

On this page