Commit Graph
2725 Commits
Author SHA1 Message Date
Willie 3b8d495153 Merge pull request #464 from runkids/runkids/458-opencode-env
fix(doctor): warn about OpenCode overlap only when skills leak
2026-10-07 13:50:59 +08:00
Willie 0b0240904a Merge pull request #465 from runkids/runkids/mcp-source-boundary
fix(mcp): keep project MCP source writes inside the project
2026-10-07 13:36:06 +08:00
Willie 920b18510a fix(doctor): count only folders with SKILL.md as hand-kept skills
An empty or cache folder in claude's skills folder was reported as a
skill OpenCode loads, and any same-named path in OpenCode's folder hid
a real one. OpenCode loads only folders with SKILL.md, so check for it
on both sides.

Refs #458
2026-10-07 13:07:25 +08:00
Willie 2ec925b880 fix(mcp): treat skillshare/config.yaml as a project config too
configProjectRoot matched only <root>/.skillshare/config.yaml, so a
project that keeps its config in the visible skillshare/ folder was
treated as a global config and its MCP saves skipped the project
boundary entirely, config.yaml included. Match every project config
folder name through projectdir.IsName.
2026-10-07 13:04:58 +08:00
Willie cc2d07d7db fix(mcp): keep project MCP source writes inside the project
A project config can name any YAML file in sources.mcp. A cloned
repository could point it at a file outside the project, such as the
global config or the global MCP source, and the next project-mode MCP
change or 0.23.0 migration rewrote that file. Saves now check the
external source against the project root, as config.yaml already is.
Reading a shared file outside the project still works.

The check also needs the right root. Deriving it from the external
file's path would accept a sibling of the project, so the root is now
passed in. A global config acting for a project in mcp.projects is no
longer treated as a project config, so a dotfile link to it keeps
working instead of being refused.
2026-10-07 13:01:31 +08:00
Willie 3ec60a3732 fix(doctor): count skills kept by hand in another target's folder
The leak check compared only source skills, so a skill the user keeps
in ~/.claude/skills (which merge mode preserves) passed as harmless,
although OpenCode loads it. Read the writer's folder too and count
entries sync did not put there and OpenCode's own folder lacks.

The wording no longer says these skills are left out by filters, and
the suggestion asks to sync them to opencode, which fits both causes.

Refs #458
2026-10-07 12:58:22 +08:00
Willie 2481179424 Merge branch 'main' into runkids/458-opencode-env 2026-10-07 12:52:39 +08:00
Willie 448a120f3e test(doctor): check cross-target discovery against a real OpenCode
Unit tests encode our reading of OpenCode's loader. This runbook
installs opencode-ai and compares `opencode debug skill` with doctor's
cross_target_discovery result, so a change in OpenCode's discovery or
env switches shows up as a failing step instead of a wrong warning.

Refs #458
2026-10-07 12:41:46 +08:00
Willie 4bca097f17 refactor(targets): write also_scans as one entry per scanned folder
disabled_by_env keyed env switches by path, so OpenCode's lists were
written twice (global and project) and every key had to match a path
listed above it, guarded only by a test. Each entry now holds a
folder's global path, project path and off_when_env together.

The loader expands entries into the same global/project lists as
before; a dump of all 79 targets is identical, order included.

Refs #458
2026-10-07 12:38:13 +08:00
Willie c3c9e06f44 docs(doctor): note when the OpenCode leak check falls back
A symlink-mode writer exposes the whole source, so doctor cannot tell
which skills OpenCode receives from that folder and keeps the general
warning. The page said the check only lists leaked skills, which is
not true in that setup.

Refs #458
2026-10-07 12:15:28 +08:00
Willie 20807579a1 refactor(doctor): keep OpenCode's one-skill-per-name fact in targets.yaml
The rule that OpenCode loads one skill per name lived in a Go map
while its other facts (also_scans, disabled_by_env) live in
targets.yaml. Move it there as one_skill_per_name so a newly verified
runtime needs only a data change.

Also fold doctor's two cross-target warning branches into one, so the
leak and generic messages cannot drift apart, and reuse cmp.Or for the
symlink-mode check.

Refs #458
2026-10-07 11:46:19 +08:00
Willie 8606dc4940 feat(doctor): honor OpenCode switches that turn off external skills
With OPENCODE_DISABLE_CLAUDE_CODE_SKILLS set, OpenCode does not read
.claude/skills, yet doctor still warned that skills leak in through
it, so a user who had already fixed the leak could never clear the
warning.

Record in targets.yaml which environment variables turn off each
also_scans folder and skip those folders in RuntimeScanPaths when a
variable is true in skillshare's environment, read the way OpenCode
reads it. doctor, sync, the dashboard and readsFrom then agree, and
doctor says which variable caused the skip, because OpenCode may run
with another environment. The doctor hint now comes from the same
table.

Refs #458
2026-10-07 11:40:51 +08:00
Willie 211efa0ee6 fix(doctor): suggest the OpenCode switch that covers the leaking folder
OPENCODE_DISABLE_CLAUDE_CODE_SKILLS only skips .claude/skills. When
skills leak through .agents/skills, for example from a universal
target, offering it as an option left OpenCode loading them after the
user followed the advice. Offer only OPENCODE_DISABLE_EXTERNAL_SKILLS
whenever .agents/skills is involved.

Refs #458
2026-10-07 11:22:49 +08:00
Willie fc2ce7d861 Merge pull request #463 from runkids/runkids/project-config-symlink-policy
fix(config): refuse project config writes that leave the project
2026-10-07 11:18:50 +08:00
Willie 03cde2b098 fix(doctor): keep the overlap warning for symlink-mode writers
A symlink-mode target folder is the source itself, so it also exposes
skills that .skillignore hides. Doctor, sync, and the dashboard pass
discovery results without those skills, so a symlink-mode claude
target looked harmless to OpenCode even when OpenCode loaded an
ignored skill through it. Treat that pairing as unknown and keep the
previous warning, unless OpenCode links the source as well.

Refs #458
2026-10-07 11:16:30 +08:00
Willie 834f88efb9 fix(doctor): compare resolved skills when judging OpenCode overlap
OpenCode loads skill files concurrently, so which copy of a name wins
is not fixed; its own folder does not reliably win. A source skill
synced to both folders is harmless anyway because both copies have
the same content. Count only source skills in OpenCode's folder, so a
local skill there no longer marks a same-named skill from another
folder as already present.

The comparison used each target's filtered selection, which skips the
target_naming: standard validation and collision handling sync applies.
When two skills named dev collide in OpenCode's standard folder, sync
writes neither there, yet a claude target filtered to one of them
still delivered it to OpenCode and no warning appeared. Use the
resolved skills that sync writes instead.

Refs #458
2026-10-07 11:10:30 +08:00
Willie d055f9c22a fix(doctor): warn about OpenCode overlap only when skills leak
OpenCode reads .claude/skills and .agents/skills first and its own
skills folder last, keeping the last skill of each name. A skill synced
to both opencode and claude therefore loads once, yet doctor and the
sync summary warned on every setup with both targets, with no way to
silence it.

The real effect is different: skills kept out of opencode by targets:
or include/exclude still reach it through another target's folder.
For opencode, count only those skills, name them in doctor, and point
at OpenCode's OPENCODE_DISABLE_CLAUDE_CODE_SKILLS and
OPENCODE_DISABLE_EXTERNAL_SKILLS. Other runtimes keep the old check
until their rule for same-named skills is verified.

Refs #458
2026-10-07 11:00:05 +08:00
Willie f5a5e920c8 fix(config): allow project configs at a volume root
The containment check appended a separator to the project root. For a
project at / or C:\ the root already ends with one, so the prefix became
// or C:\\ and every save of that project's config was refused as
linking outside the project.
2026-10-07 10:54:06 +08:00
Willie 11c2821350 fix(config): write the project migration through a fresh temp file
The legacy target migration wrote to the predictable path
config.yaml.tmp next to the validated destination. A cloned project
could ship that name as a symlink to a file outside the project, so
merely loading the project overwrote the outside file and the rename
then replaced the project config with the link.

Create the temp file with os.CreateTemp in the destination directory,
which never follows an existing name, and remove it if any step fails.
2026-10-07 10:46:00 +08:00
Willie 080bb3e5e4 fix(server): guard the dashboard raw config save in project mode
PUT /api/config wrote the raw editor content with os.WriteFile, which
follows a symlinked config.yaml. In project mode a link to a config
outside the project was overwritten, bypassing the boundary the other
config writers now enforce. Route this write through ConfigWritePath
and answer 400 with the link named.
2026-10-07 10:36:22 +08:00
Willie 04b6183ac9 Merge remote-tracking branch 'origin/runkids/project-config-symlink-policy' into runkids/project-config-symlink-policy 2026-10-07 10:33:36 +08:00
Willie 93967815f1 Merge branch 'main' into runkids/project-config-symlink-policy 2026-10-07 10:31:22 +08:00
Willie 3f1a4d1310 Merge remote-tracking branch 'origin/main' into runkids/project-config-symlink-policy 2026-10-07 10:30:47 +08:00
Willie 658e40a07b Merge pull request #462 from runkids/runkids/460-plugin-config-write
fix(plugin): keep symlinked config.yaml and normalize plugin sources
2026-10-07 10:30:06 +08:00
Willie 503247b004 Merge branch 'main' into runkids/460-plugin-config-write 2026-10-07 10:29:56 +08:00
Willie 1f277af118 fix(config): resolve the whole path before checking project containment
ConfigWritePath followed only one hop of a dangling link and resolved
only an existing parent directory. A project link to an in-project
link pointing outside, or to a missing file under an in-project
directory link that leaves the project, passed the containment check,
and the save then wrote or created the file outside the project.

Follow every link of a dangling chain, with a hop limit, and resolve
symlinks in the longest existing prefix of each path so a missing
directory cannot hide an escaping link above it.
2026-10-07 10:28:32 +08:00
Willie f4c77d53c3 fix(config): refuse project config writes that leave the project
Saves follow a symlinked config.yaml to its target so dotfile links
survive. For a project config that also lets a cloned repository point
.skillshare/config.yaml at any file the user can write, including one
that does not exist yet, and every project save, plugin change, hooks
change, MCP change, or legacy migration would then rewrite or create it.

All five writers now resolve the destination through one helper,
utils.ConfigWritePath. A global config keeps writing through its link.
A project config must resolve inside its project; otherwise the save
fails and names the link so the user can replace it with a regular
file. The one-time legacy migration skips persisting instead, and the
next save reports the link.
2026-10-07 10:19:12 +08:00
Willie f1c7c5d184 fix(plugin): expand every folded tilde form and keep dangling config links
FoldHomePathWith writes ~\plug on Windows and a bare ~ for the home
directory itself, but canonicalSource only expanded ~/, so with
preserve_tilde_on_save those sources came back relative on the next
load and update or sync could not resolve them.

When config.yaml is a symlink whose target does not exist yet, as a
dotfiles manager may create before first use, EvalSymlinks fails and
the save renamed over the link again. Follow the link to its intended
target so the first save creates that file, as Config.Save already does.

Refs #460
2026-10-07 10:02:56 +08:00
Willie 49b43b228c fix(plugin): keep symlinked config.yaml and normalize plugin sources
Plugin saves wrote a temp file next to config.yaml and renamed it over
the path. When config.yaml is a symlink into a dotfiles folder, the
rename replaced the link with a regular file, so the shared config
never received the change and the machine silently forked. The plugin
writer also skipped preserve_tilde_on_save, saving sources as absolute
home paths. The one-time legacy target migrations for global and
project configs used the same temp+rename and replaced links as well.

Plugin sources were compared as raw strings. The snapshot owner marker
records the absolute path from add, so a config that spells the same
folder as ~/... failed every update with "snapshot ownership conflict".
For Agents whose snapshot location hashes the source, the new spelling
silently created a second snapshot instead.

Resolve the symlink before writing, normalize sources once where the
plugin section is decoded so owner checks and snapshot paths agree,
and fold sources back to ~ on save when preserve_tilde_on_save is set.

Refs #460
Refs #461
2026-10-07 09:56:48 +08:00
Willie 0429a801bb Merge pull request #459 from runkids/runkids/433-submodule-detect
fix(install): refuse git submodule paths instead of installing empty dirs
2026-10-07 09:44:12 +08:00
Willie f2f663689b fix(install): warn about submodules on tracked updates
The CLI and dashboard update a tracked repo through update.TrackedRepo,
which never looked for submodules. A pull that turned a skill directory
into a submodule reported success and the skill vanished on the next
sync. TrackedRepoResult now carries the skipped-submodule warnings and
both callers show them.

Refs #433
2026-10-07 09:35:59 +08:00
Willie 0a484083e5 fix(install): warn about submodules on direct whole-repo installs
CLI search results install a repo-root skill through installFromGit without discovery, so the skipped-submodule warning has to come from that path too.
2026-10-07 02:46:26 +08:00
Willie 79bb10658a fix(install): strip submodule URL userinfo before query and fragment
A ? or # inside the userinfo used to cut away the @ before it was found, leaving the credential in place.
2026-10-07 02:40:48 +08:00
Willie b1b5bcc375 fix(install): fail closed when submodules cannot be listed and warn on tracked updates
A failed or timed-out git ls-tree used to look like a repo without submodules, letting an empty submodule path through; the refusal check now reports the failure while warnings stay best effort. Tracked updates also warn about submodules in the pulled tree.
2026-10-07 02:36:44 +08:00
Willie 6607b82e2b fix(install): escape submodule diagnostics and match ancestors by filesystem identity
Gitlink paths and .gitmodules URLs come from an untrusted repo, so quote them with %q to keep newlines and terminal escapes out of CLI output. Compare each ancestor of the subdir with the gitlink directory by os.SameFile instead of a byte-length prefix, so case and Unicode-normalization variants that open the same directory are caught.
2026-10-07 02:29:48 +08:00
Willie 39a6241b33 fix(install): refuse tracked submodule paths and fold case only where the filesystem does
A --track install with an explicit kind skips discovery, so check the subdir after the tracked clone and remove it on refusal. Match a case variant of a gitlink path only when it opens the same directory, so Linux keeps a real Vendor/Up next to a vendor/up gitlink. Strip everything before the last @ in a submodule URL so malformed and scheme-relative values fail closed.
2026-10-07 02:22:10 +08:00
Willie f1900251d3 fix(install): match submodule paths case-insensitively and warn on search installs
On Windows and macOS a case variant such as Vendor/Up opens the empty submodule checkout, so compare gitlink paths with EqualFold. Search results that install straight from discovery now show its warnings too.
2026-10-07 02:15:21 +08:00
Willie cdac248e96 fix(install): strip secrets from any submodule URL and show warnings in the dashboard
Redact query, fragment and userinfo from relative, scp-style and malformed .gitmodules URLs too, clean the subdir before matching gitlinks so vendor//up is caught, and return discovery warnings from /api/discover so the install dialog shows skipped submodules.
2026-10-07 02:10:03 +08:00
Willie 1971e9dfc4 test(install): compare submodule URL with forward slashes on Windows
git writes the .gitmodules file URL with forward slashes, so a backslash temp path never matched.
2026-10-07 02:04:59 +08:00
Willie e4b72cb554 fix(install): hide credentials in submodule URLs shown to the user
.gitmodules may hold an authenticated URL; drop userinfo, query and fragment before it reaches warnings, errors or API responses.
2026-10-07 02:03:26 +08:00
Willie b7265d6496 fix(install): refuse git submodule paths instead of installing empty dirs
skillshare never fetches submodules, so a gitlink is an empty directory
after any clone. Installing that path reported success with nothing in
it, a path inside it failed with a misleading "does not exist", and
whole-repo discovery dropped its skills without a word.

Refuse a subdir that is or lies inside a submodule, naming the
submodule, its pinned commit and upstream URL, and warn for each
submodule skipped by whole-repo discovery and --track.

Refs #433

Release-As: 0.25.1
2026-10-07 01:58:59 +08:00
Willie 3ad4b68962 Merge pull request #451 from runkids/runkids/436-managed-instruction-block
feat(extras): prepend and append modes for single-file extras
2026-10-07 01:42:09 +08:00
Willie 9ebee9b442 Merge remote-tracking branch 'origin/main' into runkids/436-managed-instruction-block
# Conflicts:
#	ui/src/components/instructions/ProjectSharedFiles.tsx
#	ui/src/components/instructions/SharedInstructions.tsx
2026-10-07 01:35:22 +08:00
Willie e881987beb Merge pull request #456 from runkids/runkids/arch-dashboard
refactor(ui): named invalidation events, shared review dialog, plugins flow module, i18n key checks
2026-10-07 01:30:08 +08:00
Willie 748352e571 Merge branch 'main' into runkids/arch-dashboard 2026-10-07 01:22:49 +08:00
Willie 06ddca6689 Merge remote-tracking branch 'origin/main' into runkids/436-managed-instruction-block 2026-10-07 01:22:15 +08:00
Willie ec17ab71e3 fix(ui): keep the floating action bar on one line
The bar is position: fixed with left: 50%, so its shrink-to-fit width is
capped at the space right of its left edge, half the viewport. Content
wider than that wrapped inside the fixed 46px height, as seen on the Pi
and OMP pending-changes bars. Size it to its content and cap it at the
viewport instead.
2026-10-07 01:14:12 +08:00
Willie 0117161c8b Merge remote-tracking branch 'origin/main' into runkids/436-managed-instruction-block 2026-10-07 01:13:44 +08:00
Willie 696e9fd1b7 Merge pull request #455 from runkids/runkids/arch-442-uninstall
refactor(uninstall): run the CLI and dashboard through one operation
2026-10-07 01:13:17 +08:00
Willie fe4fcba122 Merge branch 'main' into runkids/arch-442-uninstall 2026-10-07 01:07:38 +08:00