Registers tasks, labels, teams, task_comments and task_assignees through
the MCP tool surface, completing the v1 resource list from the plan:
* tasks : create / read_one / update / delete (read_all omitted;
models.Task.ReadAll is a stub — TaskCollection is OOS)
* labels : full CRUD
* teams : full CRUD
* tasks_comments : full CRUD, install-time gated on
config.ServiceEnableTaskComments
* tasks_assignees : create / read_all / delete only (REST exposes no
read_one or update)
Per-resource input wrappers carry the path-param fields (task_id,
user_id) explicitly so MCP callers can provide them as JSON args.
installToolsForToken fans out to one installer per resource; the
generics-bound addTool keeps per-(resource, op) call sites at compile
time. The api_tokens.yml fixture extends token 11 to cover the new
scopes; token count stays at 5 for user 1 so existing token-listing
tests are unaffected.
Integration tests per resource cover tools/list visibility, at least
one successful create or read_all, and a permission denial scenario.
Filter MCP tool visibility and invocation by the requesting API token's
(group, permission) scopes. tools/list now returns only the tools the
token's APIPermissions authorise; tools/call additionally re-checks the
scope in the dispatcher as defence-in-depth, so a session created with
one token cannot be reused to invoke tools that token never had access to.
The per-session filter runs at session-init via the StreamableHTTPHandler
getServer factory (which the SDK calls once per session, before caching
the *mcp.Server). The dispatcher check runs on every tools/call and
returns ErrScopeDenied, which the AddTool wrapper renders as an IsError
tool result.
Wires the projects resource into the MCP server end-to-end. The five
project tools (create, read_one, read_all, update, delete) are now
visible in tools/list and dispatch through handler.Do* like the REST
layer.
- Add ProjectCreateInput / ProjectUpdateInput in inputs.go with
jsonschema tags covering only the writable fields the model honours
(title, description, identifier, hex_color, parent_project_id,
position, is_archived, is_favorite); computed fields like Owner and
MaxPermission are intentionally absent so the SDK-reflected schema
stays narrow.
- Add resources.go with a sync.Once-guarded RegisterResources(), and an
installTools helper that registers tools per (resource, op) on the
*mcp.Server via a generic addTool[In inputAdapter] helper. The
handler maps domain failures (permission denials, missing rows,
validation) to IsError tool results per the SDK convention.
- Add DispatchTyped in dispatcher.go so the AddTool handler can hand a
pre-unmarshalled wrapper to the dispatcher without a JSON
round-trip. The existing Dispatch (raw JSON path) delegates to a
shared dispatchPrepared.
- Wire RegisterResources() + installTools() into newServer() so each
new MCP session inherits the static tool set.
- Add fixture token 11 (mcp:access + projects:*) for the full-scope
integration tests; bump TestAPIToken_ReadAll's expected count.
- Refresh TestMCP_ToolsListEmpty into
TestMCP_ToolsListReturnsRegisteredResources, asserting the five
projects_* tools are present (Task 6 will introduce scope-based
filtering of this list).
- Add pkg/webtests/mcp_projects_test.go covering tools/list,
create/read_one/read_all/update/delete happy paths, schema-validation
failure on missing required title, permission denial on a forbidden
project, and nonexistent-id lookup.
Define the Op bitmask, the Resource struct, the package-level Register
function, and the Dispatch entry point that future tasks will use to
expose CRUD resources over MCP. No resources are registered yet.
Op carries the CRUD-op identity, knows its api-token permission string
(matching apiTokenRoutes exactly), and knows its tool-name suffix.
Resource.Inputs maps each enabled op to a pointer-to-zero of the wrapper
type the dispatcher will allocate and unmarshal into. Register validates
the resource shape and populates a tool-name lookup table so the
dispatcher never has to string-parse names like task_comments_read_all.
Dispatch threads the user from ctx, allocates a fresh wrapper, unmarshals
arguments, asks the wrapper to copy itself onto a fresh model via the
inputAdapter seam (which Task 4 will populate with real implementations),
and forwards to the corresponding handler.Do* function. The Do* calls go
through a swappable crudFuncs struct so the unit tests can verify
dispatch routing without standing up the database.
Mount /api/v1/mcp (and /api/v1/mcp/*) inside the authenticated route
group. Reject JWT-authed requests with 401 (token-only policy), reject
API tokens without the mcp:access scope with 403, and propagate the
authed *user.User + *models.APIToken to r.Context() via typed keys so
downstream tool handlers can pull them out without depending on Echo.
The MCP protocol — JSON-RPC framing, Mcp-Session-Id management, SSE
streaming — is delegated to github.com/modelcontextprotocol/go-sdk
v1.6.1. tools/list returns {"tools": []} since no tools are registered
yet.
Adds the mcp scope group with a single access permission so it shows up
in GET /api/v1/routes (and therefore in the frontend token form).
Adds APIToken.HasMCPAccess() mirroring the caldav/feeds helpers.
The MCP endpoint will use POST, GET, and DELETE on the same path for the
streamable-HTTP transport, which CanDoAPIRoute's exact (method, path)
match cannot gate. The token middleware therefore skips the route check
for /api/v1/mcp and any sub-path; the actual authorization is delegated
to an inline HasMCPAccess() call in the MCP handler (added in the next
task).
Fixtures gain two MCP tokens for user 1: one mcp-only and one with
mcp:access plus projects read scopes for the per-tool scope filter tests.
The conversational mail template does not reference cid:logo.png, but
RenderMail still attached the embedded logo to every outgoing mail.
That left an orphan inline part that some clients render as a stray
attachment. Only embed logo.png when the formal template is in use.
Hardcoding the three exact strings localhost / 127.0.0.1 / ::1 rejected
legitimate loopback redirects like 127.0.0.2:1234 (anywhere in 127.0.0.0/8)
or [0:0:0:0:0:0:0:1]:1234 (expanded IPv6 loopback). Use net.IP.IsLoopback()
to cover the full loopback ranges, and match "localhost" case-insensitively.
0.0.0.0 stays rejected as it is not a loopback address.
https://claude.ai/code/session_01LsTDrCJ7trE6WQ4FYf78UB
Previously the OAuth server rejected every redirect_uri that did not start
with a vikunja- custom scheme. Native apps that cannot register a custom
scheme (e.g. CLIs, desktop tools) need loopback redirects per RFC 8252, so
also allow http://localhost, http://127.0.0.1 and http://[::1] (any port).
Non-loopback http:// and https:// targets remain rejected.
https://claude.ai/code/session_01LsTDrCJ7trE6WQ4FYf78UB
TaskAttachment.ReadOne now swallows ErrAccountDisabled/ErrAccountLocked
from the creator lookup, matching the existing ErrUserDoesNotExist
swallow. Without this, deleting a disabled user that owned a project
with task attachments would fail when the cascade re-loaded the
attachment to delete it.
Removes the `service.enablebotusers` config flag, the matching
`bot_users_enabled` field on /info, and the now-unused
`ErrBotUsersDisabled` error. Bot user routes and the frontend
settings tab are now always available.
https://claude.ai/code/session_01VhAR6xnoCdG1fpX52bzaCC
User search previously filtered bots only when they happened to match the
search string. That produced two bad behaviours:
1. Bots owned by other users could surface on an exact-username match,
leaking them into assignee pickers and similar UI.
2. A user could not reliably find their own bots by typing a partial
name, so bots became awkward to assign to tasks.
Change ListUsers to treat bot ownership explicitly: the existing match
branch excludes rows owned by someone else, and a second branch always
returns bots owned by the calling user. The own-bots branch also
respects any AdditionalCond passed in so project-scoped listings don't
start leaking bots from outside the project.
Bot users now render with a cool-toned (blue/cyan/violet/teal/indigo)
marble variant so they're visually distinguishable from human users.
Marble's rendering logic is parameterized with a palette; the route
forces the bot palette whenever the resolved user is a bot, overriding
whatever avatar provider they'd otherwise inherit.
The JWT skipper bypassed validation entirely for /token/test when the
bearer was an API token, leaving "user" unset in the context. CheckToken
then type-asserted it to *jwt.Token and panicked.
Validate the API token in the skipper but skip the route permission
check (since /token/test is not exposed in the API token route registry,
no token can hold explicit permission for it). Drop the now-redundant
JWT assertion in CheckToken — auth has already passed by the time the
handler runs.
Remove five keys from pkg/i18n/lang/en.json that are no longer
referenced by any i18n.T / i18n.TP call. These surfaced once the
translation check started reporting dead keys. The sibling translation
files will be reconciled on the next Crowdin sync.
Removed keys:
- notifications.task.comment.mentioned_message
- notifications.task.mentioned.message
- notifications.common.actions.assigned_you
- notifications.common.actions.assigned_themselves
- notifications.common.actions.assigned_user
The call to i18n.T for notifications.task.overdue.overdue was missing
its first positional argument, so the translation key was being passed
as the language code. This surfaced as a "dead key" once the
translation check learned to look for unused entries. Fix the call so
the reminder line is properly localised.
Keeps the Do* helpers framework-neutral so non-Echo callers (upcoming
Huma /v2 handlers) don't need a translation shim.
Addresses review feedback on #2670.
GuardLastAdmin counted only active, non-deletion-scheduled admins, but gated only on target.IsAdmin. Demoting or deleting an already-disabled or deletion-scheduled admin would then be blocked whenever exactly one active admin remained, even though removing a user who isn't in the reachable set can't reduce the count. Return early when the target isn't part of the counted set.
On startup, if the license server was unreachable with no usable cached status, or the server rejected the key, we only logged a warning without clearing persisted license.state. On Redis/keyvalue deployments a previous run's Licensed=true could remain active even though pro features were advertised as unavailable. Route both paths through degradeToFree so the persisted state is cleared.
The last-admin guard was only enforced in the --now branch of 'user delete'. The default scheduled path called user.RequestDeletion without the guard, letting an operator schedule deletion of the last reachable admin via the CLI; the cron flow would then confirm and execute it, violating the invariant the HTTP admin API already enforces.
kolaente
Frederick [Bot]
Tink bot
kolaente
MidoriKurage
Claude
Timh
Xela
