docs: post-1.3.0 sync — capture layout wording, scaffold names capture fetch

DOCS.md

@@ -93,7 +93,7 @@ EvalShift is four pieces. Each is released and documented independently; each ow
 | **GitHub Action** | `evalshift/evalshift-action@v0` | Runs the pipeline on PRs, pushes the run, maintains one PR comment, sets the `evalshift/regression` status. | [docs/github-action.md](docs/github-action.md), [action repo](https://github.com/evalshift/evalshift-action) | <https://www.evalshift.dev/ci-llms-full.txt> |
 | **Hosted server** | service — API `https://api.evalshift.dev`, web app `https://evalshift.dev` | Stores pushed run bundles, diffs runs across branches, serves the web app, drives PR comments and gating. | [docs/hosted.md](docs/hosted.md) | covered by the CLI reference (`push`/`bundle` contract) |
 
-Data flow is one-directional: **SDK captures → CLI runs and bundles → server stores and diffs → web app displays.** The SDK and CLI never call each other — the interface is files under `.evalshift/captures/`. The CLI (import `evalshift_cli`) depends on the SDK (import `evalshift`), so one environment holds both.
+Data flow is one-directional: **SDK captures → CLI runs and bundles → server stores and diffs → web app displays.** The SDK and CLI never call each other — the interface is the capture layout under `.evalshift/captures/`, on disk or mirrored from a bucket you own (`captures.store`). The CLI (import `evalshift_cli`) depends on the SDK (import `evalshift`), so one environment holds both.
 
 The CLI reference is generated from [llms-full.txt](llms-full.txt) at this repo's root — edit that file when CLI behaviour changes. The SDK and Action references are owned by their own repos; the copies served from `evalshift.dev` are synced from there.
 

README.md

@@ -47,8 +47,9 @@ Four pieces, released and documented independently:
 | **GitHub Action** — `evalshift/evalshift-action@v0` | Runs the pipeline on pull requests, pushes the run, posts one PR comment, sets the `evalshift/regression` status. | [docs/github-action.md](docs/github-action.md) |
 | **Hosted server** — `api.evalshift.dev`, web app at `evalshift.dev` | Optional. Stores pushed run bundles, diffs them across branches, drives PR comments and gating. | [docs/hosted.md](docs/hosted.md) |
 
-The SDK and the CLI never call each other — the interface is files under
-`.evalshift/captures/`, so either works without the other. The CLI (import
+The SDK and the CLI never call each other — the interface is the capture layout
+under `.evalshift/captures/`, on disk or mirrored from a bucket you own
+(`captures.store`), so either works without the other. The CLI (import
 package `evalshift_cli`) depends on the SDK (import name `evalshift`), so one
 environment holds both: `pip install evalshift` brings the SDK with it, and a
 production agent that only records captures installs `evalshift-sdk` alone.

src/evalshift_cli/cli/commands/_agents.py

@@ -152,7 +152,7 @@
 | `evalshift report` | Render the single-file HTML report. |
 | `evalshift compare` | Run doctor -> run -> evaluate -> analyze -> report end to end on one suite (formerly `all`). |
 | `evalshift inspect` | Inspect a run / artifact. |
-| `evalshift capture ...` | `list` / `promote` / `clean` / `diff` / `sync` captures into suites. |
+| `evalshift capture ...` | `list` / `fetch` / `promote` / `clean` / `diff` / `sync` captures into suites. |
 | `evalshift diff case` | Diff a single case between runs. |
 | `evalshift replay case` | Replay a single case. |
 | `evalshift traces import` | Import external traces. |

tests/unit/test_docs_currency.py

@@ -19,6 +19,8 @@
 import pytest
 
 import evalshift_cli
+from evalshift_cli.cli.commands._agents import AGENT_INSTRUCTIONS
+from evalshift_cli.cli.commands.capture import capture_app
 from evalshift_cli.cli.commands.init import PROVIDERS
 from evalshift_cli.models.registry import PROVIDER_ENV_VARS
 
@@ -227,3 +229,27 @@ def _whitespace_normalized(name: str) -> str:
 def test_copy_does_not_describe_the_retired_hosted_plans(term: str) -> None:
     offenders = [name for name in _PLAN_COPY_FILES if term in _whitespace_normalized(name)]
     assert offenders == [], f"{term!r} still appears in {offenders}"
+
+
+def _registered_capture_subcommands() -> list[str]:
+    """Every subcommand `evalshift capture --help` lists, straight from the Typer group."""
+    return sorted(str(cmd.name) for cmd in capture_app.registered_commands)
+
+
+@pytest.mark.parametrize("name", _registered_capture_subcommands())
+def test_scaffolded_agent_guide_names_every_capture_subcommand(name: str) -> None:
+    """`init` writes EVALSHIFT.md into user repos, and its `evalshift capture ...`
+    row is the one place a coding agent learns which subcommands exist.
+
+    `capture fetch` shipped in 1.3.0 without being added to that row, so every
+    scaffolded guide told agents the group had five subcommands. The row is
+    matched by its leading cell so a bare word like `list` elsewhere in the
+    guide cannot satisfy the check.
+    """
+    row = next(
+        line
+        for line in AGENT_INSTRUCTIONS.splitlines()
+        if line.startswith("| `evalshift capture ...`")
+    )
+
+    assert f"`{name}`" in row, f"EVALSHIFT.md's capture row does not name `{name}`"