@sidub-inc/docuoria.cli 1.0.15

package/package.json

@@ -0,0 +1,56 @@
+{
+  "name": "@sidub-inc/docuoria.cli",
+  "version": "1.0.15",
+  "description": "Install the Docuoria AI plugin into any AI tool skill directory.",
+  "keywords": [
+    "docuoria",
+    "ai-agent",
+    "skill",
+    "agentskills",
+    "pdf",
+    "extraction"
+  ],
+  "license": "MIT",
+  "author": "Sidub Inc.",
+  "homepage": "https://github.com/Sidub-Inc/Docuoria",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/Sidub-Inc/Docuoria.git",
+    "directory": "src/hosts/docuoria-cli-npm"
+  },
+  "bugs": {
+    "url": "https://github.com/Sidub-Inc/Docuoria/issues"
+  },
+  "engines": {
+    "node": ">=20.0.0",
+    "pnpm": ">=9.0.0"
+  },
+  "bin": {
+    "docuoria": "./bin/docuoria.js"
+  },
+  "main": "./dist/index.js",
+  "files": [
+    "bin/",
+    "dist/",
+    "payload/"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "scripts": {
+    "build": "tsup",
+    "test": "vitest run",
+    "prepublishOnly": "node scripts/verify-payload.mjs"
+  },
+  "dependencies": {
+    "@inquirer/prompts": "^7.0.0",
+    "commander": "^12.0.0",
+    "picocolors": "^1.1.0"
+  },
+  "devDependencies": {
+    "@types/node": "^20.0.0",
+    "tsup": "^8.0.0",
+    "typescript": "^5.4.0",
+    "vitest": "^2.0.0"
+  }
+}

package/payload/.claude-plugin/plugin.json

@@ -0,0 +1,21 @@
+{
+  "name": "docuoria",
+  "displayName": "Sidub Docuoria",
+  "version": "1.0.15",
+  "description": "Extract structured data from PDFs using template-driven pipelines. Author, classify, inspect, test, and execute extraction templates with a dotnet-script CLI surface.",
+  "author": {
+    "name": "Sidub Inc.",
+    "url": "https://github.com/Sidub-Inc"
+  },
+  "homepage": "https://github.com/Sidub-Inc/Docuoria",
+  "repository": "https://github.com/Sidub-Inc/Docuoria",
+  "license": "MIT",
+  "keywords": [
+    "pdf",
+    "extraction",
+    "pipeline",
+    "template",
+    "classification",
+    "dotnet-script"
+  ]
+}

package/payload/MANIFEST.json

@@ -0,0 +1,322 @@
+{
+  "package": "docuoria",
+  "version": "1.0.15",
+  "builtAt": "2026-06-03T02:39:22.6166355Z",
+  "sdk": {
+    "assembly": "Docuoria.dll",
+    "fileVersion": "1.0.15.0",
+    "informationalVersion": "1.0.15+Branch.main.Sha.67cad57e05c48eb1b6025c0077fa9357e32c6cdc.67cad57e05c48eb1b6025c0077fa9357e32c6cdc"
+  },
+  "files": [
+    {
+      "path": ".claude-plugin/plugin.json",
+      "sha256": "4ace45efa10dc2513678c7e262c1c0292c0ecfb67fb325657dd7ce0732ff967d",
+      "bytes": 632
+    },
+    {
+      "path": "assets/lib/Docuoria.dll",
+      "sha256": "64af912f8a90ec4ca5cb2c087f5e1e105ae2eb8213829431cafcac4789e70f6c",
+      "bytes": 287744
+    },
+    {
+      "path": "assets/schemas/template-schema.json",
+      "sha256": "157f4f6f0643232c892c5362a2aaaeb3927528d7b9222237de1e387afe3d710c",
+      "bytes": 15327
+    },
+    {
+      "path": "commands/classify.md",
+      "sha256": "5df5dabdb4d8dcc1b9666aec1510e6db98fb9ed05d8ecece11aa075cf4fd447c",
+      "bytes": 262
+    },
+    {
+      "path": "commands/diagnose.md",
+      "sha256": "11862ac6695e5196c131b9f13a5bdfa83053b8196fa7b3b562584e01016e07f6",
+      "bytes": 275
+    },
+    {
+      "path": "commands/extract.md",
+      "sha256": "f685936aa8bf5a3c8c45ce86e4ef0dcdb3996fd26f15015ed2a59212ce49e55b",
+      "bytes": 256
+    },
+    {
+      "path": "commands/inspect.md",
+      "sha256": "a70e3e4d7b06a9a1e6367af419b590005b0c5b55e7c3090bb14efd57570e6486",
+      "bytes": 265
+    },
+    {
+      "path": "commands/validate-template.md",
+      "sha256": "be1fcc943ca1545bcc2705f1d866c1f3b99c92d1d9dec447f0d64dee1d741179",
+      "bytes": 277
+    },
+    {
+      "path": "examples/01-extract-to-csv.md",
+      "sha256": "8742f36efad83be45a7b76410f90bd1e3a3c5a4889df30237cda19831fdf1b0d",
+      "bytes": 3641
+    },
+    {
+      "path": "examples/02-classify-unknown-pdf.md",
+      "sha256": "9d24dd10b75693104d19f95f0dfb98ee56deeed8ef69ace07968b27c573418e3",
+      "bytes": 4943
+    },
+    {
+      "path": "examples/03-diagnose-failed-result.md",
+      "sha256": "4b86cb4bf80460c1d483dd02b4454f858d04594bfe3c1676949b4cb5179de926",
+      "bytes": 3955
+    },
+    {
+      "path": "references/classification.md",
+      "sha256": "1ef5549515bf6e9161538d195c33abe841ec7d32a3224e989f83528fb663dbd1",
+      "bytes": 16252
+    },
+    {
+      "path": "references/decision-tree.md",
+      "sha256": "bd1522be6352b0c8df75b0abc8e18c3e785cef463abf0e75cbd88a58d36cf1fc",
+      "bytes": 5570
+    },
+    {
+      "path": "references/failure-tree.md",
+      "sha256": "38d7d33a07086391441c7cc3e8dcae52d999497196c1111013ce26eb9e58bcd0",
+      "bytes": 15857
+    },
+    {
+      "path": "references/pattern-authoring.md",
+      "sha256": "b678b2290cf0ba2a126bb03224c10c984d952475625dc7209b2d20ab04a9a2f2",
+      "bytes": 3372
+    },
+    {
+      "path": "references/patterns.md",
+      "sha256": "92c627dced78fc574f3bd729af7f21cf9ee8c7914b9ec4766c9374b634275566",
+      "bytes": 4156
+    },
+    {
+      "path": "references/privacy.md",
+      "sha256": "2448abb1623bed15f646eddc0875a4ba30a95d33cf66f2779f42b8dbf55cc0d0",
+      "bytes": 3502
+    },
+    {
+      "path": "references/scripts.md",
+      "sha256": "7b6a35582c789b6a24678139e11fc58e3670507a4d3532108d37832068a73827",
+      "bytes": 15127
+    },
+    {
+      "path": "references/template-reference.md",
+      "sha256": "6b640f07bfa3c553a6f274cbc639e5badc8a39652de04ad8de602631f1e15314",
+      "bytes": 25089
+    },
+    {
+      "path": "references/workflow.md",
+      "sha256": "084324d28143c8ae5994534f76fe6c82086da697c04fedba71c6672042af2ec4",
+      "bytes": 10509
+    },
+    {
+      "path": "scripts/_common.csx",
+      "sha256": "65eeea7c7f3b9383804bab2d4c8a4d204afa873f94c013d8a06181bf01fc07c5",
+      "bytes": 10518
+    },
+    {
+      "path": "scripts/classify.csx",
+      "sha256": "386bdae4c07f4b8a6305c7e2d77dda236fe36079ab618cc9811a1593f49b605a",
+      "bytes": 2229
+    },
+    {
+      "path": "scripts/dry-run.csx",
+      "sha256": "d89d7f7a5333dae60628ae43374a81ef001fcc79db99560619779bb414b79fa8",
+      "bytes": 3279
+    },
+    {
+      "path": "scripts/evaluate-match.csx",
+      "sha256": "3157aa515d648b913a829b069a8187e4979a6204518e164c79c60610e625ef83",
+      "bytes": 2727
+    },
+    {
+      "path": "scripts/execute.csx",
+      "sha256": "7e609a1f8e4883f4948c5da66f044ed1941f93b89170e758d85580264eefd8e7",
+      "bytes": 3309
+    },
+    {
+      "path": "scripts/inspect.csx",
+      "sha256": "ee7dd99e4e1905c8b60664c42b48c4f23aece97a6f3f73ed695cb0957c49ad1a",
+      "bytes": 1332
+    },
+    {
+      "path": "scripts/list-templates.csx",
+      "sha256": "6c89fd1646fc8408bbc7c8dad3f44deed8e01ba8d58baaa7b8fe32c23f2f1028",
+      "bytes": 1136
+    },
+    {
+      "path": "scripts/load-template.csx",
+      "sha256": "5dc873ed53176370a70b59091f45b7eeda86f085f6f635421d0e407b37d684f0",
+      "bytes": 1970
+    },
+    {
+      "path": "scripts/save-template.csx",
+      "sha256": "f4f9ff50b33a1f69e5f78af7c2122eb47fe3f84274e06831e56ac41be72cbed6",
+      "bytes": 1936
+    },
+    {
+      "path": "scripts/schema-info.csx",
+      "sha256": "2645c8d2e2e8f5aa0121f420e903288b33878ab62fd7811331ef8a2d6dfe5bf1",
+      "bytes": 2610
+    },
+    {
+      "path": "scripts/test-groups.csx",
+      "sha256": "71db4230f994a9946d3c62628aa2880562e0590262095fac6655500122e94ead",
+      "bytes": 1445
+    },
+    {
+      "path": "scripts/test-pattern.csx",
+      "sha256": "b61b15f163c54d0a5331c72dbea0b388ac8c1353d1fc628f165d4adee7900b66",
+      "bytes": 1982
+    },
+    {
+      "path": "scripts/validate-template.csx",
+      "sha256": "6c3db1bb1f7b0ceae5a546568b151fd16f80b3773cb1b840c01cc4f660ba4ab8",
+      "bytes": 1800
+    },
+    {
+      "path": "SKILL.md",
+      "sha256": "9f341be7cc9a5bacb2543700e9a51f5acfa14d52825d415b40e8ff8dc150b052",
+      "bytes": 5301
+    },
+    {
+      "path": "skill/assets/lib/Docuoria.dll",
+      "sha256": "64af912f8a90ec4ca5cb2c087f5e1e105ae2eb8213829431cafcac4789e70f6c",
+      "bytes": 287744
+    },
+    {
+      "path": "skill/assets/schemas/template-schema.json",
+      "sha256": "157f4f6f0643232c892c5362a2aaaeb3927528d7b9222237de1e387afe3d710c",
+      "bytes": 15327
+    },
+    {
+      "path": "skill/examples/01-extract-to-csv.md",
+      "sha256": "8742f36efad83be45a7b76410f90bd1e3a3c5a4889df30237cda19831fdf1b0d",
+      "bytes": 3641
+    },
+    {
+      "path": "skill/examples/02-classify-unknown-pdf.md",
+      "sha256": "9d24dd10b75693104d19f95f0dfb98ee56deeed8ef69ace07968b27c573418e3",
+      "bytes": 4943
+    },
+    {
+      "path": "skill/examples/03-diagnose-failed-result.md",
+      "sha256": "4b86cb4bf80460c1d483dd02b4454f858d04594bfe3c1676949b4cb5179de926",
+      "bytes": 3955
+    },
+    {
+      "path": "skill/references/classification.md",
+      "sha256": "1ef5549515bf6e9161538d195c33abe841ec7d32a3224e989f83528fb663dbd1",
+      "bytes": 16252
+    },
+    {
+      "path": "skill/references/decision-tree.md",
+      "sha256": "bd1522be6352b0c8df75b0abc8e18c3e785cef463abf0e75cbd88a58d36cf1fc",
+      "bytes": 5570
+    },
+    {
+      "path": "skill/references/failure-tree.md",
+      "sha256": "38d7d33a07086391441c7cc3e8dcae52d999497196c1111013ce26eb9e58bcd0",
+      "bytes": 15857
+    },
+    {
+      "path": "skill/references/pattern-authoring.md",
+      "sha256": "b678b2290cf0ba2a126bb03224c10c984d952475625dc7209b2d20ab04a9a2f2",
+      "bytes": 3372
+    },
+    {
+      "path": "skill/references/patterns.md",
+      "sha256": "92c627dced78fc574f3bd729af7f21cf9ee8c7914b9ec4766c9374b634275566",
+      "bytes": 4156
+    },
+    {
+      "path": "skill/references/privacy.md",
+      "sha256": "2448abb1623bed15f646eddc0875a4ba30a95d33cf66f2779f42b8dbf55cc0d0",
+      "bytes": 3502
+    },
+    {
+      "path": "skill/references/scripts.md",
+      "sha256": "7b6a35582c789b6a24678139e11fc58e3670507a4d3532108d37832068a73827",
+      "bytes": 15127
+    },
+    {
+      "path": "skill/references/template-reference.md",
+      "sha256": "6b640f07bfa3c553a6f274cbc639e5badc8a39652de04ad8de602631f1e15314",
+      "bytes": 25089
+    },
+    {
+      "path": "skill/references/workflow.md",
+      "sha256": "084324d28143c8ae5994534f76fe6c82086da697c04fedba71c6672042af2ec4",
+      "bytes": 10509
+    },
+    {
+      "path": "skill/scripts/_common.csx",
+      "sha256": "65eeea7c7f3b9383804bab2d4c8a4d204afa873f94c013d8a06181bf01fc07c5",
+      "bytes": 10518
+    },
+    {
+      "path": "skill/scripts/classify.csx",
+      "sha256": "386bdae4c07f4b8a6305c7e2d77dda236fe36079ab618cc9811a1593f49b605a",
+      "bytes": 2229
+    },
+    {
+      "path": "skill/scripts/dry-run.csx",
+      "sha256": "d89d7f7a5333dae60628ae43374a81ef001fcc79db99560619779bb414b79fa8",
+      "bytes": 3279
+    },
+    {
+      "path": "skill/scripts/evaluate-match.csx",
+      "sha256": "3157aa515d648b913a829b069a8187e4979a6204518e164c79c60610e625ef83",
+      "bytes": 2727
+    },
+    {
+      "path": "skill/scripts/execute.csx",
+      "sha256": "7e609a1f8e4883f4948c5da66f044ed1941f93b89170e758d85580264eefd8e7",
+      "bytes": 3309
+    },
+    {
+      "path": "skill/scripts/inspect.csx",
+      "sha256": "ee7dd99e4e1905c8b60664c42b48c4f23aece97a6f3f73ed695cb0957c49ad1a",
+      "bytes": 1332
+    },
+    {
+      "path": "skill/scripts/list-templates.csx",
+      "sha256": "6c89fd1646fc8408bbc7c8dad3f44deed8e01ba8d58baaa7b8fe32c23f2f1028",
+      "bytes": 1136
+    },
+    {
+      "path": "skill/scripts/load-template.csx",
+      "sha256": "5dc873ed53176370a70b59091f45b7eeda86f085f6f635421d0e407b37d684f0",
+      "bytes": 1970
+    },
+    {
+      "path": "skill/scripts/save-template.csx",
+      "sha256": "f4f9ff50b33a1f69e5f78af7c2122eb47fe3f84274e06831e56ac41be72cbed6",
+      "bytes": 1936
+    },
+    {
+      "path": "skill/scripts/schema-info.csx",
+      "sha256": "2645c8d2e2e8f5aa0121f420e903288b33878ab62fd7811331ef8a2d6dfe5bf1",
+      "bytes": 2610
+    },
+    {
+      "path": "skill/scripts/test-groups.csx",
+      "sha256": "71db4230f994a9946d3c62628aa2880562e0590262095fac6655500122e94ead",
+      "bytes": 1445
+    },
+    {
+      "path": "skill/scripts/test-pattern.csx",
+      "sha256": "b61b15f163c54d0a5331c72dbea0b388ac8c1353d1fc628f165d4adee7900b66",
+      "bytes": 1982
+    },
+    {
+      "path": "skill/scripts/validate-template.csx",
+      "sha256": "6c3db1bb1f7b0ceae5a546568b151fd16f80b3773cb1b840c01cc4f660ba4ab8",
+      "bytes": 1800
+    },
+    {
+      "path": "skill/SKILL.md",
+      "sha256": "9f341be7cc9a5bacb2543700e9a51f5acfa14d52825d415b40e8ff8dc150b052",
+      "bytes": 5301
+    }
+  ]
+}

package/payload/SKILL.md

@@ -0,0 +1,88 @@
+---
+name: docuoria
+description: Use this skill when working with Docuoria to extract structured data from PDFs, author or validate a template, design match rules for classification, diagnose a FailedResult or RejectedResult, select an ExtractionSource type, write or debug a regex pattern, or verify that PDF processing is local and private. Apply even when the user does not say "Docuoria" — any task involving the Docuoria CLI scripts, template JSON, or the IDocuoriaEngine API qualifies.
+license: MIT
+compatibility: Requires .NET 10 SDK and the `dotnet-script` global tool. SDK assembly (`Docuoria.dll`) is bundled under `assets/lib/`; transitive NuGet dependencies (PdfPig, Tabula, CsvHelper, pythonnet, Microsoft.Extensions.*) are resolved by `dotnet-script` at first run.
+---
+
+# Docuoria Skill
+
+## Installing this skill
+
+This skill directory was scaffolded by the Docuoria CLI. To install or update:
+
+```bash
+# npm (Node.js ≥ 20)
+npm install -g @sidub/docuoria
+docuoria init
+
+# .NET global tool
+dotnet tool install -g Docuoria.Cli
+docuoria init
+
+# Update an existing installation
+docuoria update
+
+# Check status / drift
+docuoria list-tools
+docuoria doctor
+```
+
+See `docs/cli.md` in the Docuoria repository for the full command reference.
+
+---
+
+## Invocation
+
+All scripts follow `dotnet script scripts/<name>.csx -- --<flag> <value>`, run from the skill root. The `--` separator is mandatory — without it, dotnet-script consumes the flags as its own. Positional arguments are rejected; pass `--help` to any script for its full flag list.
+
+Scripts divide into two groups:
+
+| Group | Scripts | Store flag |
+| --- | --- | --- |
+| **Store-aware** — read from or write to a template store | `classify`, `evaluate-match`, `list-templates`, `load-template`, `save-template` | `--store-path <dir>` or `--store-url <url>` |
+| **Standalone** — operate on individual PDF and/or template files | `inspect`, `test-pattern`, `test-groups`, `dry-run`, `execute`, `validate-template` | — |
+
+Store-aware scripts accept `--store-path <dir>` (local directory) or `--store-url <url>` (API endpoint) to locate templates; these flags are mutually exclusive. When omitted, `--store-path` defaults to `./templates` relative to the process working directory — since the CWD varies by environment, always pass the store location explicitly.
+
+## Workflow
+
+The pipeline runs in order; classification determines the entry point. Load `references/workflow.md` for the full step-by-step guide.
+
+1. **Classify** — match the PDF against all stored templates
+2. **Inspect** — read the engine's text extraction (when no template matches)
+3. **Test** — prove regex patterns against the engine's haystack
+4. **Build** — author the template JSON, validate classification rules and schema
+5. **Dry-run** — end-to-end extraction without output generation
+6. **Execute** — full pipeline producing CSV or JSON output
+7. **Store** — persist the template and verify it ranks correctly
+
+## Routing
+
+Consult the canonical reference before relying on memory. Each concern has a single owner.
+
+| If the agent needs to… | Load |
+| --- | --- |
+| Follow the full pipeline step-by-step | `references/workflow.md` |
+| Pick an `ExtractionSource` subtype for a field (`TextPattern`, `TableRows`, `TextAnchor`, `MetadataField`, `Fallback`) | `references/decision-tree.md` |
+| Design a discriminating `rootMatchRule` (token selection, composite architecture, structural rules, weights, thresholds) | `references/classification.md` |
+| Diagnose a `RejectedResult`, `FailedResult`, classification failure, or empty/incomplete `DryRunSucceeded` | `references/failure-tree.md` |
+| Map a stderr `error.code` to a remediation branch | `references/failure-tree.md` § Stderr error.code → Branch routing |
+| Copy a regex pattern from the library or adapt one to a specific PDF | `references/patterns.md` then `references/pattern-authoring.md` |
+| Look up a CLI script's flags, output envelope, or error codes | `references/scripts.md` |
+| Look up a template JSON property, `$kind` discriminator, enum value, or shape | `references/template-reference.md` |
+| Answer whether PDF processing is local/private | `references/privacy.md` |
+
+## Skill layout
+
+- `SKILL.md` — this router; loaded at skill activation.
+- `references/` — deep guides loaded on demand (see Routing table).
+- `scripts/` — `dotnet-script` CLI surface (`_common.csx` plus 11 verb scripts).
+- `assets/lib/Docuoria.dll` — bundled SDK assembly.
+- `assets/schemas/template-schema.json` — JSON Schema for template authoring and validation.
+- `examples/` — three worked end-to-end walkthroughs.
+
+## Gotchas
+
+- **`fieldType` in template JSON must be an integer (0–5), never a string.** The engine rejects string values with `RejectionReason.MalformedTemplate`. Enum: 0 String, 1 Number, 2 Integer, 3 Boolean, 4 Date, 5 Timestamp. Run `validate-template.csx` to catch this before dry-run.
+- **Adapt every regex to the actual PDF.** The engine's flattened text differs from the visual layout — whitespace, line breaks, and character encoding may not match what you see. Validate with `test-pattern.csx` and `inspect.csx` rather than pasting library patterns verbatim.