Proxy:
- server_tokens off, X-Powered-By stripped from upstream responses,
HSTS max-age raised to one year.
- No global X-Content-Type-Options: under nosniff, ORB (step 3.4)
blocks no-cors cross-origin responses typed text/html, e.g. tracking
beacons answered with an empty 204.
PHP (CVE-2026-109618, CVE-2026-109619):
- New role sys-svc-compose-php installs a shared ini (expose_php Off,
session.upload_progress.enabled 0, fatal_error_backtraces 0 for
PHP 8.5, interim memory_limit -1) and a PHP-FPM [www] drop-in
(pm.max_requests = PHP_FPM_MAX_REQUESTS) once per stack host,
before 04_files renders any compose file.
- The host file names carry a sha1 prefix of the controller-side
source, so a content change alters every PHP app's compose.yml and
recreates it, also on remote deploys where compose_volumes cannot
read the stack host file and hashes the path instead.
- 20 PHP roles mount the ini as a compose/swarm config into their
image's scan dir; moodle, nextcloud and shopware also mount the FPM
drop-in. Magento keeps its own pool: its php-fpm.conf has no
include= and is rendered without notify.
- Joomla's php-upload.ini now loads as zzz-joomla-uploads.ini so its
own memory_limit wins: the media API compares Content-Length with
toBytes(memory_limit) and rejects every upload at -1.
TODOs record the memory_limit revert condition and accepted
trade-offs, Akaunting on PHP 8.1, FusionDirectory on PHP 7.4 and the
public Keycloak admin console.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
upload.ini set memory_limit to WORDPRESS_MAX_UPLOAD_SIZE, the same 256M
as upload_max_filesize and post_max_size, so the memory a request may
use depended on how large an upload may be. It is now a quarter of the
service's mem_limit through lookup('service_memory_mb'), which renders
384M beside the 1.5g container. The upload sizes keep their variable.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The validation job asserts that main is fully merged into the PR head,
so the branch has to carry the current main before CI can pass.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
MAIL_PROVIDER moved to web-app-stalwart without a make docs run, so every generated README still drew Mailu as the email dependency. This is the generator output for the current tree: the root role table and each role cosmos now name Stalwart, Mailu loses its dependents, and Stalwart gains its swarm marker, the Tor dependencies and the 0..0 edges.
The hand-written cosmos of svc-net-tor-smtp gives way to the generated one, which no longer shows the SOCKS5 hop through svc-net-tor because the role does not declare that dependency.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Ansible warns that the name is already in use and carries on. Carrying on is the problem: inside the loop the original value is gone, so every lookup, template and nested include reading that name sees the loop item instead, and the task still reports success. The damage lands in a log nobody greps.
The lint collects every key declared in group_vars and in each role's vars or defaults, then reports any loop_var matching one. A denylist of known-bad names would have found application_id and missed domain, which is the same class in four more places.
All seven existing sites are marked, each with the reason read at the source rather than inferred. The three application_id loops pass the role id as the item, and the include_vars that follows loads that role's own application_id with the same value, so both meanings coincide by construction. The four domain loops hand the vhost to sys-stk-front-proxy, whose tasks read front_proxy_domain from exactly that name, so the shadowing is the interface.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fourteen roles repeated the same two lines to say they use the gateway: a
`shared` flag gated on group_names, and an `enabled` flag conjoining that gate
with LITELLM_CHAT_MODEL_SERVED. Both are now single-sourced.
`shared` becomes literal true. The flag means the provider is a shared instance
rather than one deployed per consumer, and a central gateway is always that;
the dynamic form only said "shared if deployed", which `enabled` already
decides. The 50 shared pins under `litellm:` leave meta/variants.yml with it.
The tree had already noticed: 20 of those 50 lines carried
`# nocheck: variants-static-override`, two of them with copy-pasted text about
container_backup. This narrows e5de4e085a for one service key rather than
reversing it; every other dynamic flag keeps the group_names form.
`enabled` becomes `{{ LITELLM_USABLE }}`, defined in group_vars/all/16_ai.yml.
That shorthand is only possible because is_explicit_truth learns about it.
The resolver reads a dependency out of the flag TEXT — "in group_names" in
value — so a constant hides the role id and the provider silently leaves the
static closure of every role referencing it. Measured: the old expression
resolves True, `{{ LITELLM_USABLE }}` resolves False. GROUP_GATED_CONSTANTS
names the constants whose own definition carries the gate, and the docstring
states the obligation: a SPOT of this kind that is not listed there drops its
provider without a word.
The wider fix, accepting any Jinja as dynamic, was measured and rejected: 35
other flags carry a Jinja value without the marker (MCP sidecars, javascript
and SSO gates, recaptcha), and they would all enter closures they are not in
today.
A literal shared flag is no longer dynamic, so test_dynamic_flags wants the
`# nocheck: dynamic-flag` its own message documents for legitimate exceptions;
where a marker already existed the rule list is extended rather than a second
comment added.
Behaviour is unchanged where it counts. The closure keeps litellm with the same
app counts (wordpress 26, openwebui 23, hermes 23, mediawiki 22, flowise 23),
unrelated Jinja still resolves False, and the truth table of the new enabled
form matches the old one in all four served/deployed combinations. Three roles
resolve without litellm, and `git show HEAD:` proves that predates this change:
zammad pins enabled false in variant 0 at line 39, nextcloud at line 97.
Six unit tests cover the resolver, including one asserting that every
registered constant is actually defined in group_vars, so an entry cannot
become a dead promise.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
host_header made the contract test send the public vhost as Host, and
web-app-wordpress was the only role in the repo carrying the flag. Under an
onion deployment home_url() is http, so the mu-plugin's https short circuit
does not fire and the filter falls through to comparing HTTP_HOST against
WORDPRESS_INTERNAL_HOST. That comparison sees the onion vhost and the service
name, never matches, application passwords stay unavailable, core never
validates the credential, and the transport guard answers 401. Clearnet hid
the defect because the first branch returned before the comparison was ever
reached.
Reproduced in three jobs across two repositories, both deploy modes and three
distributions, each with the same line: REJECTED authenticated initialize
answered 401 infinito_mcp_unauthorized. The split measured in the job logs is
https://blog.infinito.test against http://blog.<onion>.
The route without the spoof is already proven in the same failing job. hermes
reports [OK] web-app-wordpress and openclaw reads 3 tools, resources, prompts,
both against "url": "http://wordpress:80/wp-json/mcp/mcp-adapter-default-server".
WordPress serves the REST MCP route under Host: wordpress, and the filter
branch b260cc5604 added is what lets them through.
This completes 5e810c4a5a, which dropped both spoofed headers for Moodle and
moved the accommodation to the provider. b260cc5604 did the provider side for
WordPress and removed X-Forwarded-Proto but left the Host spoof standing, so
the test stayed green on a route no client uses while every client was
refused.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A client reaches the MCP endpoint by the provider's service name on the
container network, so the hop is plain HTTP. WordPress hands out an application
password only over SSL and, on an onion deployment, home_url() is
http://<onion>, so the gate stays shut: core never even attempts to validate the
credential and answers rest_not_logged_in. The client learns nothing about why.
Measured on a live deployment, same credential, same endpoint, the only
difference being the declaration of the scheme: 401 without it, 200 with it.
Deliberately wrong passwords produced the same rest_not_logged_in as correct
ones, which is what "never validated" looks like from outside.
The accommodation belongs on the provider's side, as 5e810c4a5a already settled
for Moodle: a probe that spoofs Host and X-Forwarded-Proto asserts the endpoint
works when spoofed and says nothing about how a real client reaches it. That is
exactly what happened here - the contract test sent the header and passed while
hermes and openclaw, which send only their Authorization, were refused in run
34699757369 on both compose and swarm.
WORDPRESS_INTERNAL_HOST is rendered from the same mcp.endpoint.service_key the
clients build their URL from, so both sides name one thing once. The filter
opens the gate for a request whose HTTP_HOST matches it, splits on ':' so a port
cannot defeat the match, and changes nothing when the variable is unset. Its
existing branches are untouched.
contract.py loses the X-Forwarded-Proto it used to send. The test now takes the
route a client takes, and can no longer be green while every client fails.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
WordPress registers /wp-json as a rewrite rule, so it exists only while the
permalink structure is non-empty. A fresh install ships an empty one and
nothing here ever set it: the role flushed rewrite rules it had never asked
for. Every request to /wp-json therefore matched no rule, left the query vars
empty, and WordPress answered the front page with HTTP 200.
Measured in run 34687966607, both compose+tor WordPress#0 jobs. The captured
response to an anonymous POST /wp-json/mcp/mcp-adapter-default-server is the
blog front page, not a JSON-RPC reply, and the page advertises its own REST
root as index.php?rest_route=/ - the form WordPress emits only for an empty
structure. Nothing in roles/, cli/ or scripts/ sets permalink_structure.
That 200 was read as two different accusations, both false.
test-mcp-guest.js took it for "the adapter answered an anonymous probe" and
registerMcpDisabledState would take it for "the surface is still listening
after the flag went false". 604843bb8 acted on the first one and hardened
rest_pre_dispatch, which the request never reaches; that guard only starts
working now.
The rewrite plumbing is already in place: openresty proxies without rewrite or
try_files (sys-svc-proxy/templates/location/html.conf.j2), so Apache is what
routed a path with no file behind it into index.php, which needs mod_rewrite
and .htaccess. No consumer depends on the plain form either - ?rest_route=
appears nowhere in the repository while six files use /wp-json.
The read follows the idiom 06_update_domain.yml already uses twice for home
and siteurl, failed_when: false included, and the write is an option update
rather than a rewrite structure --hard because the flush that file performs
runs right after through the include. Neither task can abort a deploy that
would otherwise have succeeded.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The mu-plugin header describes the split as 'the adapter authorises the transport, the callback authorises the operation'. The second half held, every ability carries infinito_mcp_may_read. The first did not: the adapter answered the initialize handshake to anyone, so the server and its protocol version were readable without a credential.
Measured in run 34653754175, the swarm WordPress#0: test-mcp-guest.js posts an anonymous initialize and expects at least 400, and the error context reads 'Expected: >= 400, Received: 200'. Its neighbour addons/mcp-adapter.spec.js posts an anonymous tools/list and passed, so the tool call was already refused and only the handshake was not.
The filter hangs on rest_pre_dispatch and refuses every route under the mcp namespace while the site cannot attribute the request. It leaves an already short-circuited response alone rather than overwriting it, scopes itself by route prefix so the rest of the REST API stays as open as it was, and reaches is_user_logged_in, which triggers determine_current_user and with it wp_validate_application_password, so a client presenting its application password passes. A WP_Error renders no jsonrpc key, which is the second thing the spec asserts.
The new unit test reads the guarded prefix out of the PHP and the endpoint out of the spec and requires the endpoint to sit under the prefix. Without it an adapter that moves namespace would leave the guard protecting nothing while both negative specs keep passing.
A positive spec was tried and reverted: handing the MCP credential to the Playwright env is forbidden by tests/lint/ansible/services/test_mcp_secret_not_in_playwright.py, because the trace travels in the CI artifact. No role probes its MCP endpoint with the credential at deploy time either, so the authenticated direction stays unproven.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
wp user application-password delete takes uuids as positional arguments and knows no --name flag, so the call ended on 'Error: Parameter errors: unknown --name parameter' with rc 1 and failed the deploy at mcp.yml:51. Run 34653754175, the compose WordPress#0.
The list now asks for --fields=uuid,name --format=json instead of --field=name, and a single fact holds the uuids carrying the wanted name. That one list answers both questions the block asks: whether a password of that name exists at all, and which ids to delete. Two expressions over the same fact could drift; one cannot.
The semantics are unchanged. The drop still runs only when the store cannot read the token back and WordPress holds a password under that name, and the mint follows it.
The fact is named wordpress_mcp_credential_uuids rather than ..._app_password_uuids because the password-quote lint matches the token 'password' anywhere in a shell expression and requires a trailing | quote, which map('quote') per element does not satisfy. Renaming was the fix; no suppression was added.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
At -vvv Ansible prints each module argument and result, so a password in
a command line, a request body, a header or a fact lands in the deploy
log. Nothing enforced masking outside the MCP task files, and no check
looked at parents at all.
A new lint, test_credential_tasks_are_masked, finds tasks whose printed
parts (module arguments, environment, loop) reference an identifier ending
in password, secret, token, api_key or private_key, or
secrets.credentials. A task passes when it or an enclosing block sets
no_log, when every file that pulls it in does so through include_tasks
with apply.no_log or import_tasks with no_log (transitively), when the
secret only reaches a parameter the module masks itself, or when the
module masks its values. no_log on an include_tasks task itself does not
count, because it does not reach the included tasks. The rule can be
suppressed per task with `# nocheck: credential-masking`, now listed in
the suppression catalog, and the lint tests its parent and suppression
logic on synthetic cases.
It found 39 unmasked tasks in 19 roles, all real: join tokens, client
secrets, API tokens in Cloudflare headers, database passwords in
command lines, the ChatGPT bot login body and more. Each now sets
no_log from MASK_CREDENTIALS_IN_LOGS. The sys-dns-hetzner-rdns alias
hetzner_no_log, which only restated that switch, is gone.
Not covered yet: template and copy tasks whose rendered file carries a
secret print it as a diff under --diff.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The CLI contract test rejected WordPress in job 102705652296 of run
34421944328:
REJECTED server serves ['mcp-adapter-discover-abilities', 'mcp-adapter-execute-ability', 'mcp-adapter-get-ability-info'] but the contract declares ['infinito-get-post', 'infinito-list-categories', 'infinito-search-posts']
There are two layers. The mcp-adapter v0.5.0 default server lists only its
three meta-tools and reaches public abilities through discover and execute; it
takes its tool list from the mcp_adapter_default_server_config filter
(includes/Servers/DefaultServerFactory.php). And the three infinito abilities
were never registered at all: the mu-plugin hooked abilities_api_init and
passed no category, while WordPress 7.1.0 core fires only
wp_abilities_api_init, refuses wp_register_ability() outside it and requires a
category (src/wp-includes/abilities-api.php:290-297,
abilities-api/class-wp-ability.php:295-297).
The mu-plugin now registers an `infinito` category on
wp_abilities_api_categories_init, registers the abilities on
wp_abilities_api_init with that category, and sets the default server's tools
to exactly the three abilities, with empty resources and prompts, at
PHP_INT_MAX. With the meta-tools gone, execute-ability no longer reaches every
public ability any other plugin registers.
The abilities return objects ({posts}, {categories}, {post}), since MCP defines
structuredContent as a JSON object. get-post now serves published posts only,
as search-posts already did; before, it returned any published type, including
customize_changeset and wp_block entries.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The root-README overview renders a Video link per role and a reader following it expects a demo, but 26 of the 62 entries addressed a channel, a playlist or a product page: youtube.com/@zammadhq, /channel/UC..., framatube.org/c/mobilizon/videos, pgadmin.org/videos. None of those can be embedded and none shows the product without a second click.
Each replacement was resolved through YouTube's oEmbed endpoint, which returns the real title and channel, so a candidate could be checked rather than assumed. That caught a supposed LibreOffice intro published by an unrelated channel, a Discourse overview belonging to a hosting vendor, and a Zoom video answering 403 with embedding disabled. Fourteen of the new entries come from the vendor itself and two are PeerTube.
The reachability half needed the same endpoint for a different reason: YouTube answers a removed video with HTTP 200 and a Video-unavailable page, so a plain fetch proves only that YouTube is up, while oEmbed answers 400. Measured both ways before relying on it.
The check is external because reachability needs live HTTP, and like its siblings it reports an unanswered probe as unverified rather than failing, so a firewalled runner cannot redden a good URL. Only the shape is asserted, because that is a repository fact.
README.md follows rather than leads: overview.py renders the Video column from this field.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
playwright.config.js collects `**/*.@(spec|test).js`, so a module named
test-<something>.js runs only when its role's playwright.spec.js requires it.
Three roles wrote theirs as standalone specs with a top-level test() instead:
the MCP guest guards of n8n and wordpress, and the OIDC login of hermes. All
three have been dead files since they were added, so every MCP surface was
believed to be probed by a hostile guest and hermes' oauth2-proxy round trip
to be covered while none of those assertions had ever executed.
Each moves to the exports.register shape its siblings use and is appended to
its aggregator. Nothing else changes: `git diff -w` on this commit is the
test.use line becoming the register wrapper plus one require per aggregator,
and no assertion is touched. The module-level
test.use({ignoreHTTPSErrors: true}) goes away because all three aggregators
already set it for the tests they register, and a test.use inside register()
would not apply anyway. registerMcpDisabledState moves inside register() so
the switched-off case comes with it, which the addon specs do not cover since
they skip when the addon is off.
Registering them also subjects them to test.beforeEach(shared.beforeEach).
That is safe here: wordpress sets a viewport and asserts env presence, n8n
does the same and clears cookies, hermes is a no-op. None authenticates, so
the guest guards stay guests.
wordpress is the only one of the three with a green job to lose, and its
status assertion is already proven there: addons/mcp-adapter.spec.js posts to
the same endpoint with maxRedirects 0 and asserts >= 400, it matches the
collector glob, and WordPress#1 passes. The permission callback runs at the
transport before the JSON-RPC method is parsed, so initialize is refused
exactly like tools/list.
A fourth dead module, openwebui's test-ollama-chat.js, gates on a service no
role renders and follows separately.
The gateway publishes only models it has a backend for and renders model_list: [] when neither ollama nor lmstudio is deployed and no OpenRouter key is set. LITELLM_CHAT_MODEL fell back to openrouter/auto without checking that key, so consumers were configured for a model the gateway does not carry and every prompt failed with a 400.
Observed in the Mattermost#1 CI job: litellm.log carries 12 occurrences of "Invalid model name passed in model=openrouter/auto" next to 400 Bad Request, and tests/test-litellm-agent.js failed on every attempt while the rest of the suite passed.
Make the fallback conditional on the key so the name is empty when nothing is served, promote LITELLM_CHAT_MODEL_SERVED from the role vars to group_vars where consumers can see it (it was the third copy of this logic and the only correct one), and switch the 14 consumers from "litellm is deployed" to "litellm serves a model". The Playwright service flag reads the same lookup, so the spec skips together with the Ansible gate. The bool filter carries weight: the folded scalar yields the string False, which Jinja treats as truthy.
A new unit test renders both expressions against Jinja and pins the invariant that a model name exists exactly when a model is served, including the case where ollama preloads embedding models only.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
supported_version and minimum_version sat in 24 roles and were read by no template, no task and no lookup. Their only consumer was the audit report version column, which turned false the moment an automated image bump ran: qdrant reported a verified version nobody had verified, two releases long. Both leave MCP_KEYS, so a role that declares them again fails the schema lint instead of carrying a field with no effect.
Requirement 025 mandated minimum_version as a MUST and 035 carried it in an example; both are corrected here, otherwise the repository would contradict its own code.
The gate then surfaced five dead docs.mattermost.com links. Mattermost moved to /administration-guide/ and /deployment-guide/, and every replacement was probed for 200 rather than guessed. The MCP page is gone with no successor, so mattermost now points at mcpserver/README.md in the Agents repository, the source rather than a rendered mirror.
Both concerns share three files - mattermost's meta/mcp.yml, requirement 025, and the generated audit.md, which derives from a source of each - so splitting them would commit an artifact that does not match its sources.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The Production block never set one. inventory provision writes only
ansible_connection for a local host, so DOMAIN_PRIMARY stayed at the
group_vars default of infinito.localhost and the deployment served a zone its
own resolver knows nothing about. CoreDNS templates answers for
INFINITO_DOMAIN_RE and nothing else, so every lookup of the deployment's own
name came back NXDOMAIN.
Mastodon is where that surfaced. tootctl accounts create validates the address
through EmailMxValidator, which resolves MX and then, for the domain itself,
A and AAAA. All six queries returned NXDOMAIN in run 33574056480, the
validator recorded email :unreachable, and the admin account was never created.
Only the guide rows were affected: the matrix rows write
DOMAIN_PRIMARY: "{{ lookup('env', 'INFINITO_DOMAIN') }}" into their host_vars
and resolve fine.
The template now carries DOMAIN=<your-domain> in both the host and the
container form and threads it into --vars, with the container form passing it
through the docker run boundary as well. The two guide runners substitute it
the way they already substitute the SSH key, so the replay deploys under
INFINITO_DOMAIN like every other CI deploy. compose_deploy.sh moves its env
load ahead of the substitutions to have the value; host_deploy.sh had no env
load at all.
A production reader is asked for a domain now instead of silently inheriting
infinito.localhost, which no real deployment wants either. The shipped default
keeps its value: sys-svc-nss documents that *.localhost resolves through the
myhostname NSS module without any zone, and host-side tasks lean on it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
An MCP client spec signs in through the client's own OIDC button, which
redirects to Keycloak. Every provider put its MCP pairing in the lean variant,
and that variant pinned sso off - so the round that exists to prove the pairing
was the one round without an identity provider. Run 33574056480 shows what that
costs: openwebui's MCP specs reached _shared.js:131 and died on "openwebui OIDC
sign-in button must be visible on the auth page".
The rule now demands two things of the variant that carries the consumers: the
provider pins sso on, and each admitted consumer, in the variant that round
actually pulls in, does not pin it off. The second half matters because a round
deploys every dependency at its own index and falls back to variant 0 only when
the dependency has fewer, so the client shipping alongside is that one and not
the client's variant 0. A role that declares no sso, or pins it to a literal
false because upstream speaks no OIDC, is exempt from a flag it cannot have.
Twenty roles moved to satisfy it. Ten already had a second sso: false pin and
only needed the flip. Six could host the pairing in variant 0, which already
has sso on, and dropped back to two variants. Six kept a third lean variant to
carry the false polarity variants-coverage wants. moodle and mattermost took
swaps instead: moodle's variant 1 is the mandated LDAP-only branch, and
mattermost was already at the three-variant cap with sso on in its variant 2.
nextcloud also turns context_agent on in that round, the ExApp that serves the
endpoint - with it off the round proved nothing it was built to prove.
The rule moves to tests/lint/ansible/services/ because it reads meta YAML and
executes nothing, which is where integration.md says such a check belongs, and
its suppression key joins the catalog it was never listed in.
nextcloud_exapp_reregister becomes nextcloud_exapp_renew: mcp-secret-masking
matches `register:` unanchored, so the substring in the old name read as a task
that prints a secret. register.yml already masks the two tasks that carry one.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The mcp-consumer-variant guard demanded three things of a provider's MCP
round: every admitted consumer enabled, each of them pinned `services: null`
so it dragged no dependencies along, and no other service left enabled. Two
of those were the rule legislating a round's contents rather than checking
that the pairing gets proven at all.
Keep the first and drop the rest. What else the round carries is the role's
own call, so the tor/litellm exemption, the datastore exemption and the
`allowed` set they fed have no reader left and go with them.
Strip the 63 `services: null` pins the rule forced across 20 variants, all
written bare as `services:`. The topic-emptying capability behind them
stays: apply_topic still normalises None to {} for the inventory payload,
and the unit tests still cover it. Only the uses and the mandate are gone.
That leaves overridden_providers scanning a repository with nothing left to
find, so its three assertions passed over an empty dict and would have
stayed green however the function broke. Rebuild them on a synthetic role
tree, which also covers the case the repository scan never held: a variant
naming a plain flag is not an override.
Removing the pins changes what deploys. Measured across all 20 variants: 18
identical, prometheus#1 grows 23 to 31 roles, moodle#2 grows 31 to 32 and
crosses the memory budget until the following commit shrinks it back.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A round deploys a provider once, so the role that pulls it in is the only place that can say how lean it has to be. A variant entry may now carry any application config topic under a service it depends on: a mapping deep-merges, so naming one key changes that key alone, and null empties the topic. Empty was the case that needed a marker of its own, because merging an empty mapping is a no-op and would leave exactly the services the round was trying to shed.
The topic vocabulary and the merge semantics live in utils.roles.applications.topics, which is also where resources.py now gets them rather than spelling the same null-or-merge out a second time. Collecting who demands what stays in the inventory tree, refuses two roles demanding different values for one provider in the same round, and skips a role's own entity entry so the existing services.<own>.users declarations are not read as overrides.
The CI set-cover learns the same fact. Its greedy cover retires a role's variant-0 row when an earlier row of another role embeds it, on the stated premise that embedding pulls the provider's variant 0. A variant that dictates that provider's config breaks the premise from the other side: whatever ran, it was not the provider as the provider declares it. Such a pair now covers nothing, so the retired row keeps its job, its artefact and its verdict.
Measured on the current tree: 20 override pairs, 41 covered rows before and after, so nothing is revoked today. The rule bites the moment an override lands in a row that covers.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The contract check ran as an Ansible task in the middle of every provider deploy, and the two client checks ran the same way. A test that is part of the deploy cannot be run without deploying, cannot be re-run on its own, and reports its verdict as a deploy failure. Twenty-seven task files lose 491 lines to this.
It is now what it always was. Each provider ships files/test/test.sh, templates/test.env.j2 and meta/tests.yml, and test-e2e-cli stages one shared harness beside it: run.sh drives contract.sh in a throwaway container on the provider's own network, then isolation.sh for adapter-fronted providers. The read probe each provider offers moved from the calling task into meta/mcp.yml next to the allowlist it must satisfy, and the schema now rejects a read probe naming a tool outside that allowlist.
The bearer the test presents is the one a client resolves through mcp.credential, via the new mcp_client_bearer lookup over resolve_credential, which moved to utils.roles.applications.mcp so it is the same function mcp_servers renders every client connection with. The probe presented whatever the calling role passed, which for the fifteen adapter-fronted providers is a different secret than any client holds.
Sidecar isolation is read from the engine rather than measured inside the container, so it no longer needs a delegation to the node holding the task: attachment is compared by network name, which drops the gwbridge special case, and a swarm worker says it cannot measure instead of skipping silently.
Retrying is bounded by wall clock against MCP_DEADLINE_SECONDS, which scales through the same lookup('timeout') as the harness budget it has to stay under. A flag that is neither true nor false aborts rather than reading as off.
mcp-provider-probe becomes mcp-provider-cli-test: an enabled provider must ship the test rather than include the task. mcp-probe-bearer becomes mcp-adapter-bearer, tying ADAPTER_BEARER to the declared credential key now that no probe include carries it. A new integration test renders the shared env template for every provider through Ansible's own templar, so a later change to it breaks visibly instead of poisoning twenty deploys quietly.
Hermes and OpenClaw keep their client-side proof as CLI tests of their own.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
83c1c729de gave every provider a group_names-gated services entry per MCP
client, including the clients it refuses with mcp_consumer: false. The gate
is what makes the inventory resolver plan provider and client into one round,
so 19 of the repo's 82 co-deploy pairs started a client for a pairing the
provider had already declined. flowise is a live MCP client of qdrant,
homeassistant and prometheus; the other 17 providers were bringing up its
node stack for nothing.
The entries now read enabled: false, shared: false, mcp_consumer: false, and
the variants that pinned them to true are gone. Setting those variants to
false instead is not an option: test_variants_no_static_override forbids any
variant override of a literal base value in either direction.
bond leaves with them, and a new lint holds that line: collect_role_resources
skips a literally disabled entry before it reads any resource key, so a bond
there describes a cost nobody pays. Thirteen entries unrelated to MCP carried
one and are cleared too.
test_mcp_consumer_services_entry demanded the group_names shape for every
client entry, refused ones included, and now demands the opposite there.
dynamic-flag keeps its rule and takes its own documented per-key exemption.
make docs redraws twenty graphs: the edges go, the nodes stay as
'openwebui ❌' and 'flowise ❌', so the refusal is visible rather than read as
an integration.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The Exception tag labelled prose that restates why the names carry hyphens.
The reason belongs in the history of ffdef28da, not beside the value.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
WordPress mints an application password, writes it to the token store and
every MCP client is configured with it, but nothing ever presented it to the
endpoint. credential.owner and auth_subject stayed declarations, and the
role's Playwright specs cannot close that: a browser holds no bearer, and a
run that did would capture it in its trace.
The role is also the only one with exposure: public, and its implementation
is plugin, so no adapter sidecar filters the served surface. That makes the
probe's tools/list comparison the only runtime evidence behind
mutating_tools_enabled: false; an extra registered ability or an adapter
release that grows a tool would otherwise ship silently past metadata that
still promises three read tools.
The include sits after the plugin loop rather than in tasks/utils/mcp.yml
where the credential is issued: 05_enable_plugin.yml runs the addon task
before utils/install/plugin.yml, which holds the only wp plugin activate, so
a probe run from the addon reaches an unregistered REST route and a fresh
site fails on a 404 that a second deploy makes disappear.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
enabled and shared were hand-maintained in 24 roles and wrong in both
directions. Both now resolve from the mcp_consumer edges already declared
in meta/services.yml: a server takes the consumers it admits, a client
takes the servers that admit it, and the pair renders as "'peer' in
group_names or ...", or False when no peer is admitted at all. setdefault
keeps meta/mcp.yml and meta/variants.yml as the override channel, so
pinning either key still wins.
The resolved block replaces the mapping instead of being written into it.
_build_role_base_config parks the YAML cache's own object under mcp, so an
in-place setdefault injected the derived keys into every later raw read and
made test_mcp_schema pass alone but fail in a directory run.
The derivation keys on peer presence, never on a peer's own value: a
re-entrant read under _RENDER_GUARD returns the unrendered dict.
Four lints now ask the resolver instead of reading meta/mcp.yml, which they
could not do while the key was hand-written. That is what exposed the eleven
missing hostile-path specs the previous commit closed.
Carries one change that belongs to the WordPress probe but shares a file
with this one: the tool allowlist moves to the hyphenated names the adapter
actually serves.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
allowed_consumers was twenty hand-kept lists of which fourteen were identical, so every new client meant an edit in every provider. A client now declares services.<own-entity>.mcp_consumer: true once and a provider refuses one with mcp_consumer: false, which is also what tells a deliberate exclusion apart from a forgotten entry.
The group_names-gated consumer entries give the inventory resolver the co-deploy edge meta/mcp.yml never had: one of sixty-nine matrix rounds carried an MCP client before, twenty-three role/round pairs do now. bond 0.5 keeps the clients out of round 0 so the host memory budget still holds.
BREAKING CHANGE: meta/mcp.yml no longer accepts allowed_consumers; declare services.<entity>.mcp_consumer on the client instead.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The container_address lookup does not return a container name in swarm. It
returns a shell fragment, "$(/usr/bin/resolve-container-id <stack> <service>)",
which is only a name once an interpreter evaluates the substitution. The shell
module provides one. The command module does not -- it execs the binary directly,
so the fragment arrives at container exec as literal text and the call dies with
"No such container: $(...)".
Compose hides the entire class: there the same lookup yields a bare name and
command: works. The failure exists only in swarm, and no swarm run had touched
these paths -- both were added by feature commits (b23653adf8, 0421ebeb16) that
wrote the same argv shape independently, which is the argument for a rule rather
than six edits.
Six call sites moved to the shell module: the moodle egress task, the three
moodle AI-gateway tasks, and the two wordpress AI-Engine tasks. The wordpress
pair keeps its stdin, so it uses the dict form with cmd:. Moving off argv exposes
the surrounding values to word splitting, so they now go through quote -- the
address itself must NOT, because single quotes would suppress exactly the
substitution this depends on.
test_argv_exec_needs_shell.py closes the gap that let this through.
container-exec-resolver checks that the resolver is used; it has nothing to say
about whether anything can evaluate the result. The new rule checks that other
half. It reuses the sibling's argv parser, which recognises only container/docker
in first position, so an argv list that does start an interpreter
(argv: [bash, -lc, ...], as joomla does) is correctly not matched.
Run against the tree the rule reported exactly the five sites found by hand plus
bigbluebutton, and passed over joomla and the discourse task, whose address is a
config lookup rather than the resolver. Bigbluebutton is pinned to compose via
compose_mode_force and can only ever see a bare name, so it takes a nocheck.
Five container-exec-resolver markers disappear with the argv form: under shell
the resolver is the ordinary path, not an exception.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Baserow's image mounts the MCP app on the ASGI root unconditionally, so
mcp.enabled only decides whether an endpoint record exists. The
application's own vhost kept routing /mcp/*: switched off it answered
401 Endpoint not found., and a refused credential is still a surface.
Switched on it is worse - the adapter narrows fourteen upstream tools to
four, and the publicly routed native path reaches all fourteen for
anyone holding the endpoint key. web-app-homeassistant had the same
shape, an exposure: internal surface reachable publicly. Both now 404
there and stay reachable on the container network.
The public surfaces get the edge limits the contract already declared.
Zones are written per role by the role that owns them, in the shape
web-app-prometheus uses, because one round-scoped file would drop another
application's zones on the next deploy. The rate is concurrent_requests
per second rather than an eighth limit key: a client allowed N in flight,
each bounded by timeout_seconds, cannot sustain more unless every call
returns instantly. A refusal is 429, not nginx's 503, because a client
that should back off must not read it as an outage. stream_seconds
finally binds something - the adapter holds no stream, the proxy does.
The shared location snippet's three timeouts become parameters defaulting
to the 900s they hard-coded, so no existing vhost changes.
Found by the disabled-state assertion 7d8e7e87ff added, which had never
run: its helper was never staged, so the two defects hid each other.
The audit read as prose, so a role could sit under a classification nobody had checked against its own declaration, and "blocked" could not be told apart from "nobody looked". The sections are now named after the exact values a role declares, which makes the section titles the vocabulary: one rule holds every role under its own declared classification, a second rejects a classification outside the vocabulary, and a third requires a named blocker per blocked role while forbidding a blocked role from shipping an enabled surface.
The report itself is generated from the roles' own metadata and a drift check imports the generator and compares against the committed file, which caught a stale report twice while this work was in progress. It lives under docs/contributing rather than beside the requirements, because a generated file in that directory reads like a finished criterion.
The criteria across seven requirement files now carry what they rest on: what closed them, or what blocks them and where that was measured. Two coverage limits stay stated rather than hidden: the report is built only from roles that declare an MCP block, so a role blocked without one is invisible to it, and a ticked criterion here means implemented and tested rather than proven in operation.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Every role-local MCP spec gates itself on the service flag and skips when the surface is off, so nothing ever checked that switching it off removes it. A surface still serving after the flag went false was indistinguishable from one that was never deployed: both read as a green skip. The gating helper had no inverse at all, so there was no way to assert what a disabled service must not do.
A shared assertion now probes the endpoint with both a GET and a POST while the flag is false and fails on any answer that shows something listening. It counts 401 and 403 as listening, which is the point: a refused credential still means the surface is there. The check is transport-agnostic because the roles differ in their enabled-state probes, over SSE, Streamable HTTP or plain JSON-RPC, while an endpoint that is gone answers none of them. Each role passes its endpoint URL as a thunk, since some resolve it only once a test is running.
The coverage lint requires the registration, and derives the role set from meta/mcp.yml rather than from a filename convention, which is how the tenth role surfaced: its spec is not named like the other nine.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Same shape as 31fa1e5540 for the Zammad apply scripts.
A mu-plugin is included by the WordPress loader, not imported: it exports
nothing, and defined( 'ABSPATH' ) || exit ends it outside a booted site. Its
content is three wp_register_ability calls behind the abilities_api_init hook,
whose callbacks stand on get_posts and current_user_can. A unit test would
mock all of it and prove nothing.
The suppression is structurally honest but leaves the three abilities, the
permission check among them, without automated coverage. A probe against the
running instance would be the real substitute and does not exist yet.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Twenty-five specs called the Playwright APIs with default or hardcoded
timeouts. Routed over Tor those budgets are too small, so a spec fails on
transport latency rather than on the behaviour it asserts.
Route the waits through resolveTimeout and reach pages via gotoOnion, the
helpers the tor branch introduced for exactly this. The remaining changes in
the diff are import wrapping around those calls.
The repository lint enforces this: a raw numeric timeout in a role spec is
rejected with a pointer to resolveTimeout, which is how a hardcoded 500ms
wait added during this session was caught.
Merge work rather than authored here; not deployed.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The task that renders msmtprc.conf.j2 into the WordPress config did not set
sys_svc_mail_msmtp_in_container. That variable is defined only by
sys-svc-mail-msmtp/tasks/enable.yml, so rendering the template without the
mail role failed the whole deploy with
'sys_svc_mail_msmtp_in_container' is undefined
true is the right value: WordPress runs msmtp in its own container, and its
healthcheck only writes the mail marker after msmtp exits 0.
This restores a setting rather than adding one. d0516716c5 put the same
variable on wordpress in tasks/main.yml and on moodle as a task-level var; a
later refactor dropped the wordpress line. It is declared here on the
template task itself, matching the moodle placement from that commit.
The gap stayed hidden because it only surfaces when the mail role is absent
from the round.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The ai-engine addon declared the custom environment with an id, endpoint and
api key but no models array. AI Engine resolves the models of a custom
environment in core.php through get_engine_models(), which iterates
$engine['models'], so an absent key yields an empty list and the requested
model can never be resolved. POST mwai/v1/ai/completions answered 500 before
any request left WordPress: the gateway log showed only 200s from other
consumers for that window.
Declare the single model the gateway serves, in the field shape the plugin's
own constants/models.php uses. Price and unit are zero because the
in-cluster gateway is not metered.
The two nocheck directives are needed because the addon schema lint reads a
bare integer as a possibly inlined secret; these are context windows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
