diff --git a/CHANGELOG.md b/CHANGELOG.md
index 0bab864..9e2b862 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,22 @@ All notable changes to this project will be documented in this file.
 Format follows [Keep a Changelog](https://keepachangelog.com/). This project
 adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.4.1] - 2026-07-10
+
+### Added
+
+- `explain` now resolves the console-script name (`fleet`) as well as the distribution name (`fleet-cli`); both key the root catalog entry.
+- Regression test `test_explain_self_name_matches_console_script` pins that both spellings and the bare root render identically.
+
+### Changed
+
+- `CLAUDE.md` expanded from the scaffold seed into full repo guidance: agent-first CLI contracts, the noun-group specialization checklist, test-module map, and the `fleet` vs `fleet-cli` naming split.
+
+### Fixed
+
+- The agent-first rubric gate (`teken cli doctor . --strict`) now passes 26/26; it was failing `explain_self` and turning the CI `lint` job red on every PR.
+- The `explain` remediation hint pointed at `fleet-cli explain fleet-cli`, which is not a runnable command; it now names `fleet explain fleet`.
+
 ## [0.4.0] - 2026-06-23
 
 ### Added
diff --git a/CLAUDE.md b/CLAUDE.md
index 13707ff..f216844 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -1,28 +1,265 @@
-# CLAUDE.md — seed / bootstrap placeholder
+# CLAUDE.md
 
-> **This is a self-initializing seed, not a finished runtime prompt.**
-> Run `/init` (or describe the agent's domain to your AI assistant) to
-> re-initialize this file into a full runtime prompt, using the description
-> below and the scaffolded repo as context.
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
 
-## Agent
+## What this repo actually is
 
-This repository hosts the **fleet-cli** agent.
+`fleet-cli` is an **AgentCulture mesh agent**, scaffolded from the
+`culture-agent-template`. Two things are true at once, and you must hold both:
 
-## Description
+- **Stated domain (the goal):** "Agent and CLI to control a fleet of drones
+  (multiple UAVs), built on `drone-cli` — coordinate, command, and monitor many
+  drones at once." This is the agent's identity and the direction of future work.
+- **Actual code today (the reality):** a generic **agent-first CLI scaffold** —
+  identity/introspection verbs (`whoami`, `learn`, `explain`, `overview`,
+  `doctor`, `cli`) plus the vendored AgentCulture skill kit. **No drone / UAV /
+  `drone-cli` functionality exists yet.** The CLI's own self-description still
+  reads "a clonable template for AgentCulture mesh agents" (see `learn.py`,
+  `explain/catalog.py`, `README.md`).
 
-Agent and CLI to control a fleet of drones (multiple UAVs), built on drone-cli — coordinate, command, and monitor many drones at once.
+When you add the drone-fleet functionality, you are *specializing the scaffold*:
+add new noun groups under `fleet/cli/_commands/`, give them catalog entries, and
+update the self-description strings — all while preserving the contracts below.
+Do not invent drone behavior that isn't asked for, and don't assume any drone
+plumbing is present.
 
-## Re-init instruction
+## Commands
 
-This file is a seed. To expand it into your full runtime prompt:
+```bash
+uv sync                                   # install deps into .venv (uses uv.lock)
 
-1. Open this repo in Claude Code (or your preferred AI assistant).
-2. Run `/init` — the assistant will read the repo, incorporate the description
-   above, and replace this seed with a complete `CLAUDE.md`.
-3. Commit the result.
+# Run the CLI — the console script is `fleet`, NOT `fleet-cli` (see gotcha below)
+uv run fleet whoami
+uv run python -m fleet whoami             # equivalent
+uv run fleet learn --json                 # every verb supports --json
 
-Until you run `/init`, `fleet-cli` satisfies the `steward doctor`
-`prompt-file-present` and `backend-consistency` invariants (a `CLAUDE.md`
-exists and `culture.yaml` declares `backend: claude`) but the prompt is not
-yet tailored to this agent's domain.
+# Tests
+uv run pytest -n auto                     # full suite, parallel (xdist)
+uv run pytest tests/test_cli.py::test_whoami_text   # a single test
+uv run pytest -n auto --cov=fleet --cov-report=term # with coverage (CI gate: fail_under=60)
+
+# Lint / format (exactly what the CI `lint` job runs)
+uv run black --check fleet tests          # line length 100
+uv run isort --check-only fleet tests     # black profile
+uv run flake8 fleet tests                 # max-line-length 100, ignores E203/W503
+uv run bandit -c pyproject.toml -r fleet
+markdownlint-cli2 "**/*.md" "#node_modules" "#.local" "#.claude/skills" "#.teken"
+uv run teken cli doctor . --strict        # the agent-first rubric gate (see below)
+```
+
+Black/isort: drop `--check`/`--check-only` to auto-format.
+
+Two toolchain wrinkles: `markdownlint-cli2` is an **npm** tool, not a `dev`
+dependency — CI does `npm install -g markdownlint-cli2@0.21.0`, so `uv run` will
+not find it. And **flake8 is configured in `.flake8`, not `pyproject.toml`**
+(black/isort/bandit/pytest/coverage all live in `pyproject.toml`).
+
+## Critical gotcha: `fleet` vs `fleet-cli`
+
+These names are split, and the docs are inconsistent about it:
+
+- **`fleet-cli`** = the distribution / PyPI name (`pyproject.toml [project].name`),
+  the `prog=` display name in `--help`, and the string used throughout doc text,
+  the `explain` catalog, and `learn` output.
+- **`fleet`** = the **actual runnable command** — the only `[project.scripts]`
+  entry (`fleet = "fleet.cli:main"`) — and the Python package directory.
+
+So `uv run fleet-cli whoami` (as `README.md`'s quickstart and several SKILL docs
+claim) **fails** — use `uv run fleet …` or `uv run python -m fleet …`.
+
+The split is *reconciled where it is load-bearing*, not eliminated. `explain`
+accepts both spellings (`("fleet",)` and `("fleet-cli",)` both key the root
+entry), so the rubric's `explain_self` bundle passes. But `prog=` is still
+`fleet-cli`, `--help` and `learn` still say `fleet-cli`, and `README.md:22-23`
+still prints commands that don't run. Fully collapsing the two names would mean
+changing `prog=`, which several tests pin (`usage: fleet-cli`, `subject ==
+"fleet-cli"`, `"# fleet-cli"`); that is a deliberate, separate change.
+
+## The agent-first CLI architecture
+
+The whole CLI is a small, deliberate set of contracts spread across a few files.
+Read these together before changing dispatch, errors, or output:
+
+- **`fleet/cli/__init__.py` — entry + dispatch.** `main(argv)` builds the
+  argparse tree and dispatches. Two non-obvious mechanisms live here:
+  - `_CliArgumentParser` overrides argparse's `.error()` so *parse-time* errors
+    (unknown verb, bad flag) route through the same structured `error:` / `hint:`
+    contract as runtime errors and exit `1` — not argparse's default stderr/exit-2.
+    The subparsers are built with `parser_class=_CliArgumentParser` so this
+    propagates to every noun.
+  - Parse-time errors happen before `args.json` exists, so JSON-mode rendering of
+    them relies on a class-level `_json_hint` that `main()` pre-populates by
+    scanning raw `argv` for `--json`.
+  - `_dispatch` wraps *any* unexpected exception into a `CliError` so **no Python
+    traceback ever leaks to stderr**.
+- **`fleet/cli/_errors.py` — error + exit-code policy (stable contract).** Every
+  failure raises `CliError(code, message, remediation)`. Exit codes: `0` success,
+  `1` user-input error, `2` environment/setup error, `3+` reserved. Centralised
+  here; don't invent ad-hoc exits.
+- **`fleet/cli/_output.py` — stream split (stable contract).** Results → stdout,
+  errors/diagnostics → stderr, **never mixed**. JSON mode routes structured
+  payloads to the same streams. Text-mode errors render as `error: <msg>` +
+  `hint: <remediation>` (the `hint:` prefix is required by the rubric).
+- **`fleet/cli/_commands/` — one module per verb/noun.** Each exposes
+  `register(sub)` and is wired in `_build_parser`. Add a new noun group the same
+  way (there's a commented template in `_build_parser`). A noun that has
+  action-verbs **must** also expose an `overview` verb (rubric requirement — see
+  `cli.py`, which exists only to satisfy `overview_cli_noun_exists`).
+- **`fleet/explain/` — the `explain` catalog.** `catalog.py` holds `ENTRIES`, a
+  dict keyed by command-path tuples (`()` and `("fleet-cli",)` both map to root).
+  `resolve()` raises `CliError` for unknown paths. **Every registered noun/verb
+  should have a catalog entry**, and `tests/test_cli.py::test_every_catalog_path_resolves`
+  enforces that every key resolves.
+- **`whoami.py` / `doctor.py` — identity, parsed without a YAML library.**
+  `find_culture_yaml()` walks up from `__file__` (the agent's *own* identity, not
+  the caller's CWD) and `read_agent_fields()` hand-parses the first agent block.
+  This is deliberate: see the empty-runtime-deps invariant below. `doctor.py`
+  reuses both and checks the mesh invariants.
+
+The two test modules split along the same seam, and both call `main(argv)`
+in-process (no subprocess), asserting on `capsys`:
+
+- **`tests/test_cli.py`** — the core contracts: `whoami`/`learn`/`explain`, and
+  `test_every_catalog_path_resolves`, which iterates `ENTRIES` and resolves each
+  key.
+- **`tests/test_cli_introspection.py`** — mirrors the `teken` rubric in pytest, so
+  the rubric bundles fail fast locally: `overview` graceful on a bogus target,
+  `cli overview` JSON shape, the bare `cli` noun printing something, `doctor`'s
+  `{id, passed, severity, message, remediation}` per-check keys, and — the one
+  that catches most regressions — `test_cli_overview_unknown_flag_structured_error`,
+  which asserts a parse error exits `1` with `error:` + `hint:` on stderr rather
+  than argparse's default exit `2`.
+
+Practical consequence: **changing dispatch, `_output.py`, or `_errors.py` breaks
+`test_cli_introspection.py` before it breaks the rubric gate.** Run pytest first;
+it is far faster than `teken cli doctor`.
+
+## Adding a noun group (the specialization path)
+
+This is the shape of essentially all future work here — e.g. a `drone` or
+`mission` noun. Five places, in order:
+
+> **`fleet` is a reserved noun name.** `("fleet",)` keys the *root* `explain`
+> entry, because the rubric's `explain_self` bundle probes `explain <console
+> script>` and the console script is `fleet`. A `fleet` noun group would have to
+> overwrite that key to document itself, which re-breaks the gate. The conflict
+> is inherent to the rubric, not to the catalog's lookup scheme — no aliasing
+> mechanism can make `explain fleet` render both the root docs and a noun's docs.
+> Name the drone-fleet noun something else (`drone`, `swarm`, `mission`).
+
+1. **Create `fleet/cli/_commands/<noun>.py`** exposing `register(sub) -> None`.
+   Copy `cli.py`; it is the smallest complete example.
+2. **Wire it in `_build_parser`** (`fleet/cli/__init__.py`) — import it and call
+   `_<noun>_group.register(sub)`. There is a commented template at the call site.
+   Always build the noun's parser from the passed-in `sub.add_parser(...)`: that
+   makes it a `_CliArgumentParser`, and `add_subparsers` then does
+   `kwargs.setdefault('parser_class', type(self))`, so the structured-error
+   contract propagates to nested verbs automatically. (`cli.py:40` passes
+   `parser_class=type(p)` explicitly — that is belt-and-braces, not required.)
+   What *would* break the contract is constructing a parser with a bare
+   `argparse.ArgumentParser`. `test_cli_overview_unknown_flag_structured_error`
+   is the guard.
+3. **Give the noun an `overview` verb** if it carries action verbs. `overview.py`
+   exports `emit_overview()` / `render_text()` for exactly this reuse, and the
+   rubric's `overview_cli_noun_exists` bundle is what `cli.py` exists to satisfy.
+4. **Add `explain` catalog entries** in `fleet/explain/catalog.py` — one for
+   `("<noun>",)` and one per `("<noun>", "<verb>")`.
+5. **Handlers return `int | None` and raise `CliError` on failure.** Never
+   `sys.exit()`, never print to stdout directly — use `emit_result`.
+
+**Trap:** `test_every_catalog_path_resolves` walks the catalog **keys → commands**,
+not the reverse. A new noun with *no* catalog entry therefore passes the whole
+pytest suite silently; nothing fails until a human runs `explain <noun>`. Adding
+the catalog entry is on you, not on the test.
+
+## Invariants you must not break
+
+These are enforced by tests and/or the CI rubric gate — breaking one fails CI:
+
+1. **Runtime dependencies stay empty.** `pyproject.toml` has `dependencies = []`
+   and the README sells "no third-party dependencies." `teken` is **dev-only**.
+   The hand-rolled `culture.yaml` parser exists precisely to avoid a YAML runtime
+   dep — don't add `pyyaml` (or anything) to runtime deps without a deliberate
+   decision.
+2. **No traceback leaks.** All failures go through `CliError` → `emit_error`.
+3. **stdout/stderr never mix**, and **every command supports `--json`**.
+4. **Every `explain` catalog path resolves** (test) and every CLI surface has an
+   entry.
+5. **The agent-first rubric passes:** `teken cli doctor . --strict`. It checks
+   seven bundles (learn ≥200 chars mentioning purpose/command-map/exit-codes/
+   `--json`/`explain`; `explain` resolves self + root and fails-with-hint on
+   bogus; global + per-noun `overview`; descriptive verbs never hard-fail on a
+   bad target; `doctor` shape; structured errors carry hints).
+
+## The rubric gate (green — keep it that way)
+
+`uv run teken cli doctor . --strict` passes **26/26** bundles. It regressed
+historically on `explain_self`: the rubric derives the CLI's self-name from the
+console-script entry point (`fleet`) and probes `explain fleet`, while the
+catalog only keyed `("fleet-cli",)`. Fixed by registering both spellings, and
+pinned in pytest by `test_explain_self_name_matches_console_script` so the gap
+surfaces in the fast suite rather than only in the slow gate.
+
+If you rename the console script in `[project.scripts]`, add the matching catalog
+key in the same commit — nothing derives it automatically.
+
+## Mesh identity (`culture.yaml`)
+
+`culture.yaml` declares this agent: `suffix: fleet-cli`, **`backend: colleague`**
+(a served Qwen model), so the resident prompt file is **`AGENTS.colleague.md`**,
+not `CLAUDE.md`. `doctor.py`'s backend → prompt-file map is the source of truth:
+`claude → CLAUDE.md`, `colleague → AGENTS.colleague.md`, `acp → AGENTS.md`,
+`gemini → GEMINI.md`. (Note: the seed text and a few stray strings have
+historically claimed `backend: claude` — that is stale; the declared backend is
+`colleague`.)
+
+What the tests actually pin here is narrower than "the backend is `colleague`":
+`test_doctor_recognizes_declared_backend` asserts the declared backend is a
+*known* key of `_PROMPT_FILE` and that `doctor` exits healthy. So **deleting
+`AGENTS.colleague.md` fails the suite**, but flipping `culture.yaml` to
+`backend: claude` would still pass (because `CLAUDE.md` exists). If you change
+the backend, update `_PROMPT_FILE` and the prompt file together — the suite will
+not catch a half-done rename.
+
+## Version-bump-every-PR
+
+CI's `version-check` job fails any PR whose `pyproject.toml` version equals
+`main` — **including docs/config/CI-only PRs**. Bump `version` in `pyproject.toml`
+and prepend a `CHANGELOG.md` entry (Keep a Changelog format) before opening a PR;
+use the `/version-bump` skill. PyPI Trusted Publishing fires on push to `main`
+(`.github/workflows/publish.yml`); PRs publish a `.devN` build to TestPyPI.
+
+## Skills are vendored cite-don't-import
+
+`.claude/skills/` holds skills copied verbatim from upstream, not installed as
+deps. `docs/skill-sources.md` is the authoritative provenance ledger (upstream,
+origin, last-synced, and tracked local divergences).
+
+**The ledger is currently incomplete:** 14 skill directories exist on disk but
+only 12 have ledger rows — `recall` and `remember` are vendored and unrecorded.
+Don't read the ledger as exhaustive until that's reconciled. (The README's
+"canonical guildmaster skill kit (11 skills)" is a *different* count and is
+correct: exactly 11 rows have a `guildmaster` upstream; `ask-colleague` comes
+from `colleague`, and three devague skills are re-broadcast through guildmaster.)
+
+Rules:
+
+- Most skills come from **guildmaster**; `ask-colleague` is vendored **directly
+  from `colleague`** (guildmaster hasn't re-broadcast the rename) as a tracked
+  divergence.
+- **Do not reformat or edit vendored script bodies** — they're cited, not owned.
+  markdownlint and Sonar both exclude `.claude/skills/**` for this reason. The
+  only sanctioned edits are the one consumer-identifying prose clause in each
+  `SKILL.md` and adding `type: command` to frontmatter (load-bearing for the
+  culture skill loader). Re-sync via the procedure in `docs/skill-sources.md`.
+- Reach for `ask-colleague` reflexively for a diverse second opinion: `review` a
+  committed diff before a PR, `explore` an unfamiliar area — both are read-only
+  (isolated worktree, no side effects). Side-effecting `write --apply`/`--pr`
+  needs the user's go-ahead.
+
+## Quality gate
+
+SonarCloud (`agentculture_fleet-cli`) gates the `test` job when `SONAR_TOKEN` is
+set (`sonar.qualitygate.wait=true`); token-less repos and fork PRs skip it and
+stay green. `coverage.xml` uses `relative_files = true` so paths map to
+`sonar.sources=fleet`. The `cicd` skill (`status`/`await`) queries this gate.
diff --git a/README.md b/README.md
index fe7c373..c503a95 100644
--- a/README.md
+++ b/README.md
@@ -19,11 +19,14 @@ Agent and CLI to control a fleet of drones (multiple UAVs), built on drone-cli 
 ```bash
 uv sync
 uv run pytest -n auto                 # run the test suite
-uv run fleet-cli whoami  # identity from culture.yaml
-uv run fleet-cli learn   # self-teaching prompt (add --json)
+uv run fleet whoami                   # identity from culture.yaml
+uv run fleet learn                    # self-teaching prompt (add --json)
 uv run teken cli doctor . --strict    # the agent-first rubric gate CI runs
 ```
 
+The console script is **`fleet`** — `fleet-cli` is the distribution name on PyPI,
+not a runnable command. `uv run python -m fleet …` works too.
+
 ## CLI
 
 | Verb | What it does |
diff --git a/fleet/explain/__init__.py b/fleet/explain/__init__.py
index 1cc8b77..a698dde 100644
--- a/fleet/explain/__init__.py
+++ b/fleet/explain/__init__.py
@@ -16,7 +16,10 @@ def resolve(path: tuple[str, ...]) -> str:
     raise CliError(
         code=EXIT_USER_ERROR,
         message=f"no explain entry for: {display}",
-        remediation="list entries with: fleet-cli explain fleet-cli",
+        # Name the console script, not the distribution: `fleet-cli` is not a
+        # runnable command, so a hint spelling it that way sends the reader to
+        # a "command not found".
+        remediation="list entries with: fleet explain fleet",
     )
 
 
diff --git a/fleet/explain/catalog.py b/fleet/explain/catalog.py
index 72700ea..7be9e10 100644
--- a/fleet/explain/catalog.py
+++ b/fleet/explain/catalog.py
@@ -1,7 +1,11 @@
 """Markdown catalog for ``fleet-cli explain <path>``.
 
 Each entry is verbatim markdown. Keys are command-path tuples. The empty tuple
-and ``("fleet-cli",)`` both resolve to the root entry.
+resolves to the root entry, as do both spellings of this CLI's name: the
+distribution name ``("fleet-cli",)`` and the console-script name ``("fleet",)``.
+Both are registered because the agent-first rubric derives the CLI's self-name
+from the ``[project.scripts]`` entry point (``fleet``) and probes
+``explain fleet``, while the doc text throughout addresses it as ``fleet-cli``.
 
 Keep bodies self-contained: an agent reading one entry should get enough
 context without chaining reads.
@@ -118,7 +122,12 @@
 
 ENTRIES: dict[tuple[str, ...], str] = {
     (): _ROOT,
-    ("fleet-cli",): _ROOT,
+    ("fleet-cli",): _ROOT,  # distribution name (prog=, doc text)
+    # Console-script name. RESERVED: `resolve()` does exact-tuple lookup, so this
+    # key cannot also document a noun group named `fleet` — and the rubric's
+    # `explain_self` bundle requires it to render the root entry. Name any future
+    # noun `drone`/`swarm`/`mission`, never `fleet`. See CLAUDE.md.
+    ("fleet",): _ROOT,
     ("whoami",): _WHOAMI,
     ("learn",): _LEARN,
     ("explain",): _EXPLAIN,
diff --git a/pyproject.toml b/pyproject.toml
index a6016f8..d05fb49 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "fleet-cli"
-version = "0.4.0"
+version = "0.4.1"
 description = "Agent and CLI to control a fleet of drones (multiple UAVs), built on drone-cli — coordinate, command, and monitor many drones at once."
 readme = "README.md"
 license = "Apache-2.0"
diff --git a/tests/test_cli.py b/tests/test_cli.py
index 4ab0ee2..1795d4c 100644
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -113,3 +113,21 @@ def test_every_catalog_path_resolves(capsys: pytest.CaptureFixture[str]) -> None
         rc = main(["explain", *path])
         assert rc == 0, f"explain {' '.join(path)} failed"
         capsys.readouterr()
+
+
+def test_explain_self_name_matches_console_script(capsys: pytest.CaptureFixture[str]) -> None:
+    """Both spellings of this CLI's name resolve to the root entry.
+
+    The agent-first rubric derives the CLI's self-name from the
+    ``[project.scripts]`` entry point (``fleet``) and probes ``explain fleet``,
+    but the doc text addresses the CLI as ``fleet-cli``. Registering only the
+    latter fails the rubric's ``explain_self`` bundle. Pinned here so the gap
+    surfaces in pytest (fast) rather than only in `teken cli doctor` (slow).
+    """
+    assert main(["explain"]) == 0
+    root = capsys.readouterr().out
+
+    for name in ("fleet", "fleet-cli"):
+        rc = main(["explain", name])
+        assert rc == 0, f"explain {name} failed"
+        assert capsys.readouterr().out == root, f"explain {name} != bare root"
diff --git a/uv.lock b/uv.lock
index 47aaca5..2a80fd7 100644
--- a/uv.lock
+++ b/uv.lock
@@ -154,9 +154,32 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/61/e8/cb8e80d6f9f55b99588625062822bf946cf03ed06315df4bd8397f5632a1/coverage-7.14.0-py3-none-any.whl", hash = "sha256:8de5b61163aee3d05c8a2beab6f47913df7981dad1baf82c414d99158c286ab1", size = 211764, upload-time = "2026-05-10T18:02:29.538Z" },
 ]
 
+[[package]]
+name = "execnet"
+version = "2.1.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/bf/89/780e11f9588d9e7128a3f87788354c7946a9cbb1401ad38a48c4db9a4f07/execnet-2.1.2.tar.gz", hash = "sha256:63d83bfdd9a23e35b9c6a3261412324f964c2ec8dcd8d3c6916ee9373e0befcd", size = 166622, upload-time = "2025-11-12T09:56:37.75Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ab/84/02fc1827e8cdded4aa65baef11296a9bbe595c474f0d6d758af082d849fd/execnet-2.1.2-py3-none-any.whl", hash = "sha256:67fba928dd5a544b783f6056f449e5e3931a5c378b128bc18501f7ea79e296ec", size = 40708, upload-time = "2025-11-12T09:56:36.333Z" },
+]
+
+[[package]]
+name = "flake8"
+version = "7.3.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "mccabe" },
+    { name = "pycodestyle" },
+    { name = "pyflakes" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/9b/af/fbfe3c4b5a657d79e5c47a2827a362f9e1b763336a52f926126aa6dc7123/flake8-7.3.0.tar.gz", hash = "sha256:fe044858146b9fc69b551a4b490d69cf960fcb78ad1edcb84e7fbb1b4a8e3872", size = 48326, upload-time = "2025-06-20T19:31:35.838Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/9f/56/13ab06b4f93ca7cac71078fbe37fcea175d3216f31f85c3168a6bbd0bb9a/flake8-7.3.0-py2.py3-none-any.whl", hash = "sha256:b9696257b9ce8beb888cdbe31cf885c90d31928fe202be0889a7cdafad32f01e", size = 57922, upload-time = "2025-06-20T19:31:34.425Z" },
+]
+
 [[package]]
 name = "fleet-cli"
-version = "0.3.4"
+version = "0.4.1"
 source = { editable = "." }
 
 [package.dev-dependencies]
@@ -185,29 +208,6 @@ dev = [
     { name = "teken", specifier = ">=0.8" },
 ]
 
-[[package]]
-name = "execnet"
-version = "2.1.2"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/bf/89/780e11f9588d9e7128a3f87788354c7946a9cbb1401ad38a48c4db9a4f07/execnet-2.1.2.tar.gz", hash = "sha256:63d83bfdd9a23e35b9c6a3261412324f964c2ec8dcd8d3c6916ee9373e0befcd", size = 166622, upload-time = "2025-11-12T09:56:37.75Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/ab/84/02fc1827e8cdded4aa65baef11296a9bbe595c474f0d6d758af082d849fd/execnet-2.1.2-py3-none-any.whl", hash = "sha256:67fba928dd5a544b783f6056f449e5e3931a5c378b128bc18501f7ea79e296ec", size = 40708, upload-time = "2025-11-12T09:56:36.333Z" },
-]
-
-[[package]]
-name = "flake8"
-version = "7.3.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "mccabe" },
-    { name = "pycodestyle" },
-    { name = "pyflakes" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/9b/af/fbfe3c4b5a657d79e5c47a2827a362f9e1b763336a52f926126aa6dc7123/flake8-7.3.0.tar.gz", hash = "sha256:fe044858146b9fc69b551a4b490d69cf960fcb78ad1edcb84e7fbb1b4a8e3872", size = 48326, upload-time = "2025-06-20T19:31:35.838Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/9f/56/13ab06b4f93ca7cac71078fbe37fcea175d3216f31f85c3168a6bbd0bb9a/flake8-7.3.0-py2.py3-none-any.whl", hash = "sha256:b9696257b9ce8beb888cdbe31cf885c90d31928fe202be0889a7cdafad32f01e", size = 57922, upload-time = "2025-06-20T19:31:34.425Z" },
-]
-
 [[package]]
 name = "iniconfig"
 version = "2.3.0"