ragfly-cli 1.18.0__tar.gz → 1.19.1__tar.gz

ragfly_cli-1.19.1/PKG-INFO

@@ -0,0 +1,127 @@
+Metadata-Version: 2.4
+Name: ragfly-cli
+Version: 1.19.1
+Summary: RAGfly CLI — operate RAGfly from the terminal and CI (login + cloud API). Lightweight, no desktop dependencies.
+Author: CAB Ltda
+License: MIT
+Project-URL: Homepage, https://ragfly.ai
+Project-URL: Documentation, https://api.ragfly.ai/docs
+Keywords: ragfly,rag,cli,retrieval,ai,mcp
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: click>=8.1
+Requires-Dist: rich>=13.0
+Requires-Dist: httpx>=0.27
+Requires-Dist: keyring>=25.0
+Requires-Dist: pydantic>=2.0
+Requires-Dist: pydantic-settings>=2.6
+Dynamic: license-file
+
+# RAGfly CLI
+
+Operate RAGfly from the terminal and CI — `login` + the full `cloud` API surface
+against `api.ragfly.ai`. Lightweight: depends only on `click`, `rich`, `httpx`
+and `keyring`. **No desktop / PySide6 / OCR dependencies.**
+
+```bash
+pip install ragfly-cli
+ragfly version
+ragfly login
+ragfly cloud me
+```
+
+> This package ships the **`ragfly` binary** for scripting and automation.
+> It is distinct from `pip install ragfly` (the **Python SDK**, import `import ragfly`)
+> and from **RAGfly Desktop** (the DMG/exe that bundles the local file worker for
+> `ragfly local scan/sync/daemon`).
+
+## Authentication
+
+```bash
+# A person (JWT, stored in the OS keyring)
+ragfly login
+
+# Non-interactive / CI / agents (API key, only reaches /v1)
+export RAGFLY_API_KEY=rf_xxxxxxxxxx
+```
+
+`cloud group` and `cloud api-key` manage a person's session and credentials: they
+need `ragfly login`, and the server refuses them for an API key.
+
+## Opening an original file on disk (optional)
+
+Searching and citing need no local path setup. A document result may include an
+`fs` object. For a local relative file, read `fs.home_var` and join that
+environment variable's value with `fs.relative_path`. Each root has its own
+variable; configure only the roots available on this machine.
+
+```bash
+export RAGFLY_HOME_442681="/Users/ana/Dropbox"
+export RAGFLY_HOME_991203="/Volumes/Archive"
+```
+
+If `fs.is_cloud_only` is true, use the authorized cloud provider. If
+`fs.is_public_url` is true, open its public URL. If `fs.is_absolute` is true,
+use the absolute path on the machine where it exists. If `fs.home_var` is null
+or unset, do not guess a local root. See the
+[MCP guide](https://ragfly.ai/build/mcp#opening-a-document-on-disk-fs-block)
+for the full order and examples.
+
+## Command surface
+
+```
+ragfly
+├── login / logout / version
+└── cloud                      ← the English REST /v1 contract of api.ragfly.ai
+    ├── me
+    ├── group          list | switch | clear      (signed-in person)
+    ├── api-key        create | list | revoke     (signed-in person)
+    ├── document       list | show | edges
+    ├── space          list | show
+    ├── queue          show | runs
+    ├── skill          list | show | run
+    ├── catalog
+    ├── function       show
+    ├── search
+    ├── chat           ask
+    ├── agent          context | tool
+    ├── usage
+    ├── conversation   list | delete
+    ├── process        list | show | update
+    ├── organization   show | update | draft
+    └── operation      list | show | run
+```
+
+Every operation of the RAGfly application is available through `cloud operation`,
+with the same permissions and audit as the web app:
+
+```bash
+ragfly cloud operation list
+ragfly cloud operation show document_types.update          # input/output schema
+ragfly cloud operation run document_types.update --input-json '{"code": "TDOC_...", "name": "Invoices"}'
+
+# write_confirm operations (deletes, reverts, resets) only run with --confirm
+ragfly cloud operation run document_types.delete --input-json @input.json --confirm
+```
+
+`--input-json` takes inline JSON, `@file` or `-` (stdin). Never put secrets in it.
+Output is English end to end (`--status VECTORIZED`, `-o json | jq`).
+
+```bash
+ragfly cloud document list --status VECTORIZED --limit 20 -o id
+ragfly cloud skill run SUMMARIZE_DOCUMENT --space 12
+ragfly cloud search "Q1 revenue"
+```
+
+Full reference: <https://api.ragfly.ai/docs>.
+
+> **Local operations** (`ragfly local scan/sync/daemon`) require the local file
+> worker and ship with **RAGfly Desktop**, not with this package.
+
+## Relation to RAGfly Desktop
+
+RAGfly Desktop bundles its own `ragfly` binary with the local commands and a
+signed-in person's session. This package is the standalone cloud surface: since
+1.19.0 its `cloud` commands use only the public `/v1` contract, so they no longer
+mirror the Desktop copy.

ragfly_cli-1.19.1/README.md

@@ -0,0 +1,107 @@
+# RAGfly CLI
+
+Operate RAGfly from the terminal and CI — `login` + the full `cloud` API surface
+against `api.ragfly.ai`. Lightweight: depends only on `click`, `rich`, `httpx`
+and `keyring`. **No desktop / PySide6 / OCR dependencies.**
+
+```bash
+pip install ragfly-cli
+ragfly version
+ragfly login
+ragfly cloud me
+```
+
+> This package ships the **`ragfly` binary** for scripting and automation.
+> It is distinct from `pip install ragfly` (the **Python SDK**, import `import ragfly`)
+> and from **RAGfly Desktop** (the DMG/exe that bundles the local file worker for
+> `ragfly local scan/sync/daemon`).
+
+## Authentication
+
+```bash
+# A person (JWT, stored in the OS keyring)
+ragfly login
+
+# Non-interactive / CI / agents (API key, only reaches /v1)
+export RAGFLY_API_KEY=rf_xxxxxxxxxx
+```
+
+`cloud group` and `cloud api-key` manage a person's session and credentials: they
+need `ragfly login`, and the server refuses them for an API key.
+
+## Opening an original file on disk (optional)
+
+Searching and citing need no local path setup. A document result may include an
+`fs` object. For a local relative file, read `fs.home_var` and join that
+environment variable's value with `fs.relative_path`. Each root has its own
+variable; configure only the roots available on this machine.
+
+```bash
+export RAGFLY_HOME_442681="/Users/ana/Dropbox"
+export RAGFLY_HOME_991203="/Volumes/Archive"
+```
+
+If `fs.is_cloud_only` is true, use the authorized cloud provider. If
+`fs.is_public_url` is true, open its public URL. If `fs.is_absolute` is true,
+use the absolute path on the machine where it exists. If `fs.home_var` is null
+or unset, do not guess a local root. See the
+[MCP guide](https://ragfly.ai/build/mcp#opening-a-document-on-disk-fs-block)
+for the full order and examples.
+
+## Command surface
+
+```
+ragfly
+├── login / logout / version
+└── cloud                      ← the English REST /v1 contract of api.ragfly.ai
+    ├── me
+    ├── group          list | switch | clear      (signed-in person)
+    ├── api-key        create | list | revoke     (signed-in person)
+    ├── document       list | show | edges
+    ├── space          list | show
+    ├── queue          show | runs
+    ├── skill          list | show | run
+    ├── catalog
+    ├── function       show
+    ├── search
+    ├── chat           ask
+    ├── agent          context | tool
+    ├── usage
+    ├── conversation   list | delete
+    ├── process        list | show | update
+    ├── organization   show | update | draft
+    └── operation      list | show | run
+```
+
+Every operation of the RAGfly application is available through `cloud operation`,
+with the same permissions and audit as the web app:
+
+```bash
+ragfly cloud operation list
+ragfly cloud operation show document_types.update          # input/output schema
+ragfly cloud operation run document_types.update --input-json '{"code": "TDOC_...", "name": "Invoices"}'
+
+# write_confirm operations (deletes, reverts, resets) only run with --confirm
+ragfly cloud operation run document_types.delete --input-json @input.json --confirm
+```
+
+`--input-json` takes inline JSON, `@file` or `-` (stdin). Never put secrets in it.
+Output is English end to end (`--status VECTORIZED`, `-o json | jq`).
+
+```bash
+ragfly cloud document list --status VECTORIZED --limit 20 -o id
+ragfly cloud skill run SUMMARIZE_DOCUMENT --space 12
+ragfly cloud search "Q1 revenue"
+```
+
+Full reference: <https://api.ragfly.ai/docs>.
+
+> **Local operations** (`ragfly local scan/sync/daemon`) require the local file
+> worker and ship with **RAGfly Desktop**, not with this package.
+
+## Relation to RAGfly Desktop
+
+RAGfly Desktop bundles its own `ragfly` binary with the local commands and a
+signed-in person's session. This package is the standalone cloud surface: since
+1.19.0 its `cloud` commands use only the public `/v1` contract, so they no longer
+mirror the Desktop copy.

{ragfly_cli-1.18.0 → ragfly_cli-1.19.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ragfly-cli"
-version = "1.18.0"
+version = "1.19.1"
 description = "RAGfly CLI — operate RAGfly from the terminal and CI (login + cloud API). Lightweight, no desktop dependencies."
 readme = "README.md"
 requires-python = ">=3.10"

ragfly_cli-1.19.1/ragfly_cli/__init__.py

@@ -0,0 +1 @@
+__version__ = "1.19.1"

{ragfly_cli-1.18.0 → ragfly_cli-1.19.1}/ragfly_cli/_http.py

@@ -41,9 +41,8 @@ def default_headers(
     Returns:
         dict de headers listo para pasar a httpx.
 
-    El header `X-Override-Grupo` se envía cuando hay grupo activo configurado
-    o explícito. El backend lo respeta como override de sesión (mismo patrón
-    que el dropdown de grupo del frontend web).
+    El header `X-Override-Grupo` solo se envía para una sesión humana. Una API
+    key tiene un grupo fijo en el servidor y nunca hereda el grupo local del CLI.
     """
     from ragfly_cli import __version__ as _client_version
 
@@ -53,14 +52,17 @@ def default_headers(
     if content_type:
         headers["Content-Type"] = "application/json"
 
-    # Resolver grupo activo: explícito > config
-    grupo = grupo_override
-    if grupo is None:
-        try:
-            from .config import get_config
-            grupo = get_config().codigo_grupo or None
-        except Exception:
-            grupo = None
+    # El override pertenece a la sesión humana. Enviar el grupo guardado en
+    # otra sesión con una API key puede dar 403 y nunca debe cambiar su alcance.
+    grupo = None
+    if not (token and token.startswith(("rf_", "slm_live_"))):
+        grupo = grupo_override
+        if grupo is None:
+            try:
+                from .config import get_config
+                grupo = get_config().codigo_grupo or None
+            except Exception:
+                grupo = None
     if grupo:
         headers["X-Override-Grupo"] = grupo