boost-skill-cli 1.0.216__tar.gz → 1.0.217__tar.gz

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/.vale/styles/config/vocabularies/boost/accept.txt

@@ -4,6 +4,7 @@ argparse
 bmad
 bookended
 CI's
+Codecov
 codespell
 config
 configs
@@ -22,6 +23,7 @@ GIF
 gitignored
 glassmorphic
 hardcode
+hotspots
 jonnyeclectic
 jonnyeclectic's
 langchain
@@ -31,6 +33,7 @@ mergeable
 mutmut
 mypy
 npm
+onboards
 Pinecone
 pytest
 rebased
@@ -46,6 +49,7 @@ rulesets
 Rulesets
 sandboxed
 sandboxing
+SonarCloud
 stderr
 Stdlib
 stdlib

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/CONTRIBUTING.md

@@ -63,6 +63,23 @@ are pinned to the same tool versions the lint gate uses, and
 [pre-commit.ci](https://pre-commit.ci) runs them on every PR and pushes any
 auto-fixes straight back onto the branch.
 
+## Optional dashboards (Codecov, SonarCloud)
+
+Two hosted dashboards are wired up but **inert until someone onboards them**.
+Both are deliberately *non-blocking*: boost's own gates run offline and are the
+authoritative ones, and a merge should never hinge on a third party's uptime.
+
+| Dashboard | To enable | Config |
+|---|---|---|
+| **Codecov** — PR diff-coverage comments, file sunburst, trend | add a `CODECOV_TOKEN` repo secret | [`codecov.yml`](codecov.yml) |
+| **SonarCloud** — bugs, smells, security hotspots, duplication | import the repo at [sonarcloud.io](https://sonarcloud.io) (Analysis Method: GitHub Actions), then add the generated token as `SONAR_TOKEN` | [`sonar-project.properties`](sonar-project.properties) |
+
+Without the secret each step skips itself — the `if:` reads the token through
+`env`, since a secret can't be referenced directly in an `if`. The SonarCloud job
+also writes a note to the run summary saying what to do, so a skipped run
+explains itself instead of looking broken. Neither is in the required-checks
+list, and neither should be added to it.
+
 ## The gates (CI runs all of these)
 
 | Command | Gate |

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/Makefile

@@ -10,7 +10,7 @@ PYTEST    := $(VENV)/bin/pytest
 # pinned taps out of the developer's real ~/.boost.
 EVAL_HOME := $(CURDIR)/.eval-home
 
-.PHONY: venv test unit functional smoke coverage patch-coverage mutation lint check demo carousel clean-test eval eval-ai eval-rec eval-stats eval-explain audit dist-check bdd bench bench-cli fuzz
+.PHONY: venv test unit functional smoke coverage patch-coverage mutation lint check demo carousel clean-test eval eval-ai eval-rec eval-stats eval-explain audit dist-check bdd bench bench-cli fuzz post-deploy
 
 # Every tool comes from a hash-pinned requirements/*.txt — the same files CI
 # installs (see scripts/lock_toolchain.py). pip enforces the hashes, so a dev
@@ -155,6 +155,16 @@ fuzz:
 			-max_total_time=$(FUZZ_SECONDS) -artifact_prefix=/tmp/ || exit 1; \
 	done
 
+# Health check for the PUBLISHED docs site: every page 200, every local asset and
+# internal link resolves, and (with Chrome) no console errors or failed runtime
+# requests. Not part of `make check` — it needs the network and the live deploy;
+# the post-deploy workflow runs it after Pages republishes.
+# BOOST_SITE=http://localhost:8000/ make post-deploy   to check a local serve.
+BOOST_SITE ?= https://jonnyeclectic.github.io/boost/
+post-deploy:
+	$(PY) scripts/post_deploy_smoke.py --base-url "$(BOOST_SITE)" -v
+	@command -v node >/dev/null 2>&1 && BOOST_SITE="$(BOOST_SITE)" node tests/visual/console_check.mjs || echo "node not on PATH — skipping the console check (CI runs it)"
+
 # regenerate generated artifacts from their source (registries + roadmap boards)
 generate:
 	$(PY) scripts/build_registries.py

{boost_skill_cli-1.0.216/boost_skill_cli.egg-info → boost_skill_cli-1.0.217}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boost-skill-cli
-Version: 1.0.216
+Version: 1.0.217
 Summary: boost — Homebrew for AI coding skills
 Author: Jonathan Reyes
 License: GNU GENERAL PUBLIC LICENSE

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/boost_cli/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '1.0.216'
-__version_tuple__ = version_tuple = (1, 0, 216)
+__version__ = version = '1.0.217'
+__version_tuple__ = version_tuple = (1, 0, 217)
 
 __commit_id__ = commit_id = None

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/boost_cli/commands/pkg.py

@@ -109,7 +109,7 @@ def _warn_secrets(res: store.InstallResult) -> None:
         out.warn("  L%d [%s] %s" % (f.line, f.severity, f.description))
 
 
-def _offer_mcp(res: store.InstallResult) -> None:
+def _offer_mcp(res: store.InstallResult, no_mcp: bool = False) -> None:
     """Surface the MCP servers a just-installed skill needs, and offer to wire them.
 
     A skill that only works paired with an MCP server used to install "cleanly"
@@ -118,11 +118,11 @@ def _offer_mcp(res: store.InstallResult) -> None:
     servers that came with a runnable spec.
 
     Advisory, never fatal: a declined prompt, a missing ``claude`` CLI or a failed
-    registration all leave the skill installed. Suppressed entirely by
-    ``--no-mcp``, which sets ``BOOST_NO_MCP_OFFER``.
+    registration all leave the skill installed. Suppressed by ``--no-mcp`` (passed
+    down as ``no_mcp``) or by exporting ``BOOST_NO_MCP_OFFER``.
     """
     rows = res.mcp_servers
-    if not rows or os.environ.get("BOOST_NO_MCP_OFFER"):
+    if not rows or no_mcp or os.environ.get("BOOST_NO_MCP_OFFER"):
         return
     from ..core import mcpdecl
     out.info("")
@@ -170,7 +170,7 @@ def _register_mcp_server(name: str, spec: dict) -> None:
     journal.log("mcp-register", name)
 
 
-def _report_result(res: store.InstallResult) -> None:
+def _report_result(res: store.InstallResult, no_mcp: bool = False) -> None:
     """The canonical three-line install report."""
     if res.kind in ("rule", "workflow"):
         where = " (this repo)" if res.scope == "project" else ""
@@ -202,7 +202,7 @@ def _report_result(res: store.InstallResult) -> None:
     out.ok("lock updated (.skill-lock.json)")
     _warn_injection(res)
     _warn_secrets(res)
-    _offer_mcp(res)
+    _offer_mcp(res, no_mcp=no_mcp)
 
 
 def _boostfile_text(skills: Dict[str, dict], via: str = "boost bundle dump") -> str:
@@ -304,11 +304,6 @@ def cmd_install(argv: List[str]) -> int:
                     help="don't offer to register MCP servers a skill declares")
     args = ap.parse_args(argv)
     args.scope = args.scope or scopes.SCOPE_USER
-    if args.no_mcp:
-        # An env flag rather than threading a parameter through _report_result's
-        # eight call sites: the offer is a leaf-level side effect, and this keeps
-        # every other install path (update, reinstall, team, MCP tool) unchanged.
-        os.environ["BOOST_NO_MCP_OFFER"] = "1"
     only = _check_agents(args.agent)
     multi = len(args.names) > 1
     entries, failed = [], 0
@@ -394,7 +389,7 @@ def cmd_install(argv: List[str]) -> int:
             out.warn("%s: %s" % (e["name"], err.message))
             failed += 1
             continue
-        _report_result(res)
+        _report_result(res, no_mcp=args.no_mcp)
         results.append(res)
     if results:
         kinds = {r.kind for r in results}

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217/boost_skill_cli.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boost-skill-cli
-Version: 1.0.216
+Version: 1.0.217
 Summary: boost — Homebrew for AI coding skills
 Author: Jonathan Reyes
 License: GNU GENERAL PUBLIC LICENSE

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/boost_skill_cli.egg-info/SOURCES.txt

@@ -16,11 +16,13 @@ Makefile
 README.md
 SECURITY.md
 boost
+codecov.yml
 eslint.config.mjs
 index.html
 noxfile.py
 pyproject.toml
 setup.cfg
+sonar-project.properties
 .vale/styles/boost/Terminology.yml
 .vale/styles/config/vocabularies/boost/accept.txt
 boost_cli/__init__.py
@@ -254,6 +256,7 @@ docs/roadmap/items/required-checks-config-drift.md
 docs/roadmap/items/retrieval-quality-eval-harness.md
 docs/roadmap/items/reuse-helpers-kill-minor-dead-work.md
 docs/roadmap/items/robust-tag-argument-parsing.md
+docs/roadmap/items/ruff-016-widens-the-default-rule-set.md
 docs/roadmap/items/rule-install-native-materialization.md
 docs/roadmap/items/runner-egress-monitoring-stepsecurity-harden-run.md
 docs/roadmap/items/runtime-explain-faithfulness-guardrail.md
@@ -323,6 +326,7 @@ scripts/import_budget.py
 scripts/lock_toolchain.py
 scripts/mutation_gate.py
 scripts/perf_gate.py
+scripts/post_deploy_smoke.py
 style/README.md
 style/boost.css
 style/boost.js
@@ -438,6 +442,7 @@ tests/unit/test_pkg_confusions.py
 tests/unit/test_pkg_injection.py
 tests/unit/test_pkg_secrets.py
 tests/unit/test_policy.py
+tests/unit/test_post_deploy_smoke.py
 tests/unit/test_projectlock.py
 tests/unit/test_provenance.py
 tests/unit/test_rag.py
@@ -462,5 +467,6 @@ tests/unit/test_util.py
 tests/unit/test_version.py
 tests/unit/test_workflows.py
 tests/visual/a11y_check.mjs
+tests/visual/console_check.mjs
 tests/visual/package.json
 tests/visual/visual_check.mjs

boost_skill_cli-1.0.217/codecov.yml

@@ -0,0 +1,46 @@
+# Codecov configuration.
+#
+# Codecov is a DASHBOARD here, not a gate. boost already enforces coverage
+# itself, twice, and those checks are the authoritative ones:
+#
+#   * project floor  — pytest --cov with fail_under = 80 (pyproject.toml)
+#   * patch floor    — diff-cover --fail-under=80 vs origin/main (ci.yml)
+#
+# Both run in-repo, offline, on every PR. Layering a third-party status check on
+# top would mean a merge could be blocked by somebody else's uptime, and a gate
+# you cannot run locally is a gate contributors cannot preempt. So every status
+# below is `informational: true`: Codecov comments, draws the sunburst and tracks
+# the trend, and never fails a build.
+#
+# Until CODECOV_TOKEN exists as a repo secret the upload step is skipped
+# entirely (see ci.yml), so this file is inert rather than wrong.
+
+coverage:
+  precision: 1
+  round: down
+  # Match the repo's own floor so the dashboard and the gate tell the same story.
+  range: "80...100"
+  status:
+    project:
+      default:
+        target: 80%
+        informational: true
+    patch:
+      default:
+        target: 80%
+        informational: true
+
+comment:
+  # One comment, rewritten in place, showing the diff and the file breakdown.
+  layout: "condensed_header, diff, flags, components, condensed_files"
+  behavior: once
+  require_changes: true      # stay silent on a PR that changes no covered code
+
+ignore:
+  # Generated by setuptools-scm at build time and absent from a checkout — the
+  # same carve-out [tool.mypy] and [tool.pyright] make.
+  - "boost_cli/_version.py"
+  # Dev/CI tooling and fixtures, not shipped code.
+  - "scripts/**"
+  - "tests/**"
+  - "noxfile.py"

boost_skill_cli-1.0.217/docs/roadmap/items/coverage-dashboard-codecov-free-for-oss.md

@@ -0,0 +1,25 @@
+---
+id: coverage-dashboard-codecov-free-for-oss
+board: code
+section: dx
+status: shipped
+category: Testing · Platform
+complexity: S
+impact: Med
+wow: 3
+note: PR decoration
+order: 8
+owner: loop/quality-dashboards
+pr: 246
+title: Coverage dashboard — Codecov (free for OSS)
+---
+The self-contained <code>diff-cover</code> gate enforces patch coverage in CI; this is
+           its hosted counterpart — the same <code>coverage.xml</code> the gate already
+           reads, turned into PR diff-coverage comments, a file sunburst and a trend line
+           over time. Wired into the existing <code>patch-coverage</code> job (no second
+           test run) and <em>deliberately non-blocking</em>: <code>codecov.yml</code> marks
+           every status <code>informational</code>, because boost already enforces coverage
+           twice offline and a merge should never hinge on a third party's uptime. Inert
+           until a <code>CODECOV_TOKEN</code> secret exists — the step's <code>if</code>
+           reads the token through <code>env</code> (a secret cannot be referenced directly
+           in an <code>if</code>), so with no secret it skips rather than fails.

boost_skill_cli-1.0.217/docs/roadmap/items/post-deploy-smoke-headless-load-check.md

@@ -0,0 +1,28 @@
+---
+id: post-deploy-smoke-headless-load-check
+board: code
+section: docsite
+status: shipped
+category: Testing · Docs
+complexity: M
+impact: Med
+wow: 3
+note: catches red deploys
+order: 9
+owner: loop/post-deploy-smoke
+pr: 244
+title: Post-deploy smoke — headless load check
+---
+Every other docs gate runs against the working tree — <code>html-validate</code> and
+           the a11y checker parse the files, the visual sweep loads them over
+           <code>file://</code>. All of them can be green while the <em>deployed</em> site is
+           broken, because deployment is where the paths change: Pages serves this repo
+           under <code>/boost/</code>, so a link that resolves on disk can 404 in production.
+           Nothing was watching that. Now, triggered by the Pages deployment finishing (so it
+           checks what was actually published rather than guessing at propagation),
+           <code>scripts/post_deploy_smoke.py</code> asserts every page answers 200 and every
+           local asset and internal link resolves — pure stdlib over HTTP, no browser — and
+           <code>console_check.mjs</code> loads each live page in headless Chrome for the
+           half only a browser sees: uncaught JS, <code>console.error</code>, and runtime
+           request failures. Off-site links are deliberately excluded: a flaky third party
+           must never redden boost's deploy, and <code>links.yml</code> already owns those.

boost_skill_cli-1.0.217/docs/roadmap/items/quality-dashboard-sonarcloud-free-for-oss.md

@@ -0,0 +1,27 @@
+---
+id: quality-dashboard-sonarcloud-free-for-oss
+board: code
+section: health
+status: shipped
+category: Quality · Platform
+complexity: M
+impact: High
+wow: 5
+note: PR quality gate
+order: 6
+owner: loop/quality-dashboards
+pr: 246
+title: Quality dashboard — SonarCloud (free for OSS)
+---
+One external surface over everything the offline gates measure separately: bugs, code
+           smells, security hotspots, duplication and a coverage overlay, decorating each PR
+           with a quality gate. Free for public repos. Settings live in
+           <code>sonar-project.properties</code> — <code>boost_cli</code> as sources with
+           <code>tests</code> declared separately (so the suite's asserts do not score as
+           duplication), the generated version stub and boards excluded, coverage read from
+           the Cobertura report CI already emits rather than re-running the suite.
+           Non-blocking and inert until onboarded: without <code>SONAR_TOKEN</code> the job
+           skips and writes what to do into the run summary, so a skipped run explains
+           itself instead of looking broken. Not added to required checks — boost's own
+           gates stay authoritative because they run offline and a contributor can
+           reproduce them.

boost_skill_cli-1.0.217/docs/roadmap/items/ruff-016-widens-the-default-rule-set.md

@@ -0,0 +1,38 @@
+---
+id: ruff-016-widens-the-default-rule-set
+board: code
+section: health
+status: planned
+category: Toolchain
+complexity: M
+impact: Med
+wow: 2
+note: 83 errors, 43 auto-fixable
+order: 15
+owner:
+pr: 236
+title: ruff 0.16 widens the default rule set — 83 new errors on a version bump
+---
+Dependabot #236 bumps ruff <code>0.15.22 &rarr; 0.16.0</code> and the <code>ruff</code> step
+fails with <b>83 errors</b>. None of them are new code — 0.16 <b>widened the default
+selection</b>, and because <code>pyproject.toml</code> uses <code>extend-select</code>
+(which adds to the defaults rather than replacing them), every newly-defaulted family
+switched on at once. The project only ever opted into
+<code>S, B, SIM, C4, PERF, RUF, UP</code>; what fires now is
+<code>I001</code>&nbsp;(33), <code>PLW1510</code>&nbsp;(19), <code>BLE001</code>&nbsp;(9),
+<code>ISC004</code>&nbsp;(4), plus <code>TRY</code>, <code>PLR</code>, <code>PIE</code>,
+<code>DTZ</code>, <code>RET</code>, <code>PYI</code>, <code>FLY</code> and — awkwardly —
+<code>FURB167</code>, a family the config explicitly delegates to refurb via
+<code>external = ["FURB"]</code>.
+So this is a <b>policy decision, not a lint fix</b>, and there are two honest answers.
+Pin the policy: switch <code>extend-select</code> to a full <code>select = [...]</code> so
+the rule set is stated outright and a future ruff release can never widen it again — the
+bump then lands with zero code changes. Or adopt some of them: <code>I001</code> (import
+sorting) and the 43 auto-fixable errors are cheap and arguably worth having, while
+<code>PLW1510</code> (<code>subprocess.run</code> without <code>check=</code>) is 19 real
+call sites that each need a judgement about whether a non-zero exit should raise.
+<code>DTZ005/006</code> (naive <code>datetime.now()</code>) is worth a look on its own
+merits given <a href="#rel-time-tests-race-the-wall-clock">the clock-racing test</a>.
+Recommended: pin <code>select</code> first so the bump is unblocked and the gate stops
+depending on an upstream default, then adopt families deliberately in follow-ups. Until
+then #236 must not be merged — it reddens <code>ci / lint</code> on every branch.

{boost_skill_cli-1.0.216 → boost_skill_cli-1.0.217}/docs/roadmap.html

@@ -109,7 +109,7 @@
       <div class="stat"><b>16</b><span>Shipped</span></div>
       <div class="stat"><b>1</b><span>Next up</span></div>
       <div class="stat"><b>4</b><span>Planned</span></div>
-      <div class="stat"><b>145</b><span>Loop finds</span></div>
+      <div class="stat"><b>146</b><span>Loop finds</span></div>
       <!-- /ROADMAP:stats -->
     </div>
   </section>
@@ -1308,13 +1308,20 @@ improvement is real (or isn't)."</p>
       </article>
 
       <article class="cap rcard" id="quality-dashboard-sonarcloud-free-for-oss">
-        <div class="head"><span class="pill planned">Planned</span><span class="cat">Quality · Platform</span></div>
+        <div class="head"><span class="pill shipped">Shipped</span><span class="cat">Quality · Platform</span></div>
         <h3>Quality dashboard — SonarCloud (free for OSS)</h3>
-        <p>SonarCloud is free for public repos and layers bugs, code smells,
-           security hotspots, duplication and a coverage overlay into one
-           dashboard, decorating each PR with a pass/fail quality gate. The
-           single highest-signal freemium add — one external surface that unifies
-           everything else in this section.</p>
+        <p>One external surface over everything the offline gates measure separately: bugs, code
+           smells, security hotspots, duplication and a coverage overlay, decorating each PR
+           with a quality gate. Free for public repos. Settings live in
+           <code>sonar-project.properties</code> — <code>boost_cli</code> as sources with
+           <code>tests</code> declared separately (so the suite's asserts do not score as
+           duplication), the generated version stub and boards excluded, coverage read from
+           the Cobertura report CI already emits rather than re-running the suite.
+           Non-blocking and inert until onboarded: without <code>SONAR_TOKEN</code> the job
+           skips and writes what to do into the run summary, so a skipped run explains
+           itself instead of looking broken. Not added to required checks — boost's own
+           gates stay authoritative because they run offline and a contributor can
+           reproduce them.</p>
         <div class="meta">
           <span class="m">Complexity <b>M</b></span>
           <span class="m hi">Impact <b>High</b></span>
@@ -1427,6 +1434,40 @@ exactly what the gate is supposed to catch.</p>
         </div>
       </article>
 
+      <article class="cap rcard" id="ruff-016-widens-the-default-rule-set">
+        <div class="head"><span class="pill planned">Planned</span><span class="cat">Toolchain</span></div>
+        <h3>ruff 0.16 widens the default rule set — 83 new errors on a version bump</h3>
+        <p>Dependabot #236 bumps ruff <code>0.15.22 &rarr; 0.16.0</code> and the <code>ruff</code> step
+fails with <b>83 errors</b>. None of them are new code — 0.16 <b>widened the default
+selection</b>, and because <code>pyproject.toml</code> uses <code>extend-select</code>
+(which adds to the defaults rather than replacing them), every newly-defaulted family
+switched on at once. The project only ever opted into
+<code>S, B, SIM, C4, PERF, RUF, UP</code>; what fires now is
+<code>I001</code>&nbsp;(33), <code>PLW1510</code>&nbsp;(19), <code>BLE001</code>&nbsp;(9),
+<code>ISC004</code>&nbsp;(4), plus <code>TRY</code>, <code>PLR</code>, <code>PIE</code>,
+<code>DTZ</code>, <code>RET</code>, <code>PYI</code>, <code>FLY</code> and — awkwardly —
+<code>FURB167</code>, a family the config explicitly delegates to refurb via
+<code>external = ["FURB"]</code>.
+So this is a <b>policy decision, not a lint fix</b>, and there are two honest answers.
+Pin the policy: switch <code>extend-select</code> to a full <code>select = [...]</code> so
+the rule set is stated outright and a future ruff release can never widen it again — the
+bump then lands with zero code changes. Or adopt some of them: <code>I001</code> (import
+sorting) and the 43 auto-fixable errors are cheap and arguably worth having, while
+<code>PLW1510</code> (<code>subprocess.run</code> without <code>check=</code>) is 19 real
+call sites that each need a judgement about whether a non-zero exit should raise.
+<code>DTZ005/006</code> (naive <code>datetime.now()</code>) is worth a look on its own
+merits given <a href="#rel-time-tests-race-the-wall-clock">the clock-racing test</a>.
+Recommended: pin <code>select</code> first so the bump is unblocked and the gate stops
+depending on an upstream default, then adopt families deliberately in follow-ups. Until
+then #236 must not be merged — it reddens <code>ci / lint</code> on every branch.</p>
+        <div class="meta">
+          <span class="m">Complexity <b>M</b></span>
+          <span class="m">Impact <b>Med</b></span>
+          <span class="m">Wow <b class="wow">★★</b></span>
+          <span class="m">83 errors, 43 auto-fixable</span>
+        </div>
+      </article>
+
       <article class="cap rcard" id="doctor-checks-rules-and-workflows">
         <div class="head"><span class="pill shipped">Shipped</span><span class="cat">Diagnostics · Install engine</span></div>
         <h3><code>boost doctor</code> checks installed rules and workflows</h3>
@@ -1912,13 +1953,18 @@ self-merge intact. Decide before adding more concurrent loops, not after.</p>
       </article>
 
       <article class="cap rcard" id="coverage-dashboard-codecov-free-for-oss">
-        <div class="head"><span class="pill planned">Planned</span><span class="cat">Testing · Platform</span></div>
+        <div class="head"><span class="pill shipped">Shipped</span><span class="cat">Testing · Platform</span></div>
         <h3>Coverage dashboard — Codecov (free for OSS)</h3>
-        <p>Upgrade the homegrown coverage badge to Codecov's PR diff-coverage
-           comments, file sunburst and trend graph — free for public repos. The
-           hosted counterpart to the self-contained <code>diff-cover</code> gate:
-           one enforces patch coverage in CI, the other visualizes the trend over
-           time.</p>
+        <p>The self-contained <code>diff-cover</code> gate enforces patch coverage in CI; this is
+           its hosted counterpart — the same <code>coverage.xml</code> the gate already
+           reads, turned into PR diff-coverage comments, a file sunburst and a trend line
+           over time. Wired into the existing <code>patch-coverage</code> job (no second
+           test run) and <em>deliberately non-blocking</em>: <code>codecov.yml</code> marks
+           every status <code>informational</code>, because boost already enforces coverage
+           twice offline and a merge should never hinge on a third party's uptime. Inert
+           until a <code>CODECOV_TOKEN</code> secret exists — the step's <code>if</code>
+           reads the token through <code>env</code> (a secret cannot be referenced directly
+           in an <code>if</code>), so with no secret it skips rather than fails.</p>
         <div class="meta">
           <span class="m">Complexity <b>S</b></span>
           <span class="m">Impact <b>Med</b></span>
@@ -2294,13 +2340,21 @@ and the work stays visible.</p>
       </article>
 
       <article class="cap rcard" id="post-deploy-smoke-headless-load-check">
-        <div class="head"><span class="pill planned">Planned</span><span class="cat">Testing · Docs</span></div>
+        <div class="head"><span class="pill shipped">Shipped</span><span class="cat">Testing · Docs</span></div>
         <h3>Post-deploy smoke — headless load check</h3>
-        <p>After each Pages deploy, headless-load the guide and roadmap and assert
-           the essentials: HTTP 200, every asset resolves, and the console is free
-           of errors. Distinct from the planned visual-regression pass — this is a
-           cheap always-on health gate that would have caught the current broken
-           Pages deploy automatically.</p>
+        <p>Every other docs gate runs against the working tree — <code>html-validate</code> and
+           the a11y checker parse the files, the visual sweep loads them over
+           <code>file://</code>. All of them can be green while the <em>deployed</em> site is
+           broken, because deployment is where the paths change: Pages serves this repo
+           under <code>/boost/</code>, so a link that resolves on disk can 404 in production.
+           Nothing was watching that. Now, triggered by the Pages deployment finishing (so it
+           checks what was actually published rather than guessing at propagation),
+           <code>scripts/post_deploy_smoke.py</code> asserts every page answers 200 and every
+           local asset and internal link resolves — pure stdlib over HTTP, no browser — and
+           <code>console_check.mjs</code> loads each live page in headless Chrome for the
+           half only a browser sees: uncaught JS, <code>console.error</code>, and runtime
+           request failures. Off-site links are deliberately excluded: a flaky third party
+           must never redden boost's deploy, and <code>links.yml</code> already owns those.</p>
         <div class="meta">
           <span class="m">Complexity <b>M</b></span>
           <span class="m">Impact <b>Med</b></span>