Add 5 new audit rules (31 → 36 total) to close detection gaps:
- fetch-with-pipe (HIGH ×3): curl/wget piped to sh/bash/python/node/etc.
Previously only double-pipe (||) was caught by shell-chain; single
pipe (|) to interpreters — the classic RCE vector — was undetected.
- ip-address-url (MEDIUM): URLs with raw IP addresses that bypass DNS
security controls; excludes RFC 1918 and loopback ranges.
- data-uri (HIGH): data: URIs in markdown links that may embed
executable content.
fetch-with-pipe is suppressed inside markdown code fences (tutorial
context), matching existing behavior for destructive-commands et al.
Includes unit tests (25 cases), code fence suppression test, red team
TC-18 through TC-21, and audit.md documentation updates.
For testing on UTM Windows VMs. Scans common shared folder
locations and presents interactive picker when multiple found.
make build-windows SHARED=~/Downloads
mise run build:windows ~/Downloads
v0.16.0 moved update rollback logic from update.go to
update_handlers.go and changed the error message format.
Update the mutation target file and perl regex pattern so
the signal test finds the expected 2 mutation points.
Swap fast path order: sparse checkout is now fast path 1 (works for all
hosts including file://), GitHub Contents API is fast path 2 (fallback
when sparse is unavailable or fails). Also reset subdirPath on sparse
failure to ensure proper fallback chain.
Fix red team helpers to inject -g flag after subcommand, preventing
.skillshare/ auto-detection from hijacking global-mode tests.
Add blocked section assertions to TestUpdate_BatchAll_FailsOnMalicious.
InstallTrackedRepo and updateTrackedRepo had no security audit,
allowing --track to bypass all supply-chain checks. This adds
auditTrackedRepo (removes repo on block for fresh installs) and
auditTrackedRepoUpdate (git reset rollback for updates), plus
input validation, testable removeAll, and hardened error handling
across audit/install/uninstall paths.
Replace simple regex-based markdown link extraction with a full
structural parser that handles inline links with titles, autolinks,
reference-style links, HTML anchors, multiline links, and correctly
skips code fences, inline code spans, and image links.
This ensures external-link and source-repository-link audit rules
match all Markdown link syntax variants while avoiding false positives
from code examples and images.
test-redteam-signal verifies the red team suite detects intentional
code mutations. test-redteam-rules-signal verifies it fails when
critical builtin audit rules are disabled.
- firstPull: attempt merge instead of blocking when both local and
remote have skills; override merge.ff=only with -c merge.ff=false
- auditSkillByName: fall back to collectInstalledSkillPaths for
short-name resolution of nested skills (e.g. security/audit-demo-*)
- Add external-link and dangling-link demo skills to sandbox content
Sandbox (8→2): consolidate into `make playground` + `playground-down`,
add scripts/sandbox.sh wrapper for advanced ops (up/down/shell/reset/
status/logs/bare).
Test (7→5): remove test-cover and test-install from Make (still
available via scripts directly).
Dev Docker (3→2): merge dev-docker-up and dev-docker-watch into
`make dev-docker` (defaults to Compose Watch for auto-rebuild),
keep dev-docker-down.
Update all docs references: README, CLAUDE.md, quick-reference,
docker-sandbox guide, ui command page.
- Add `make sandbox-logs` / `mise run sandbox:logs` convenience target
- Unify Docker build versions via ARG: Node 24, pnpm 10.28.0, Go 1.25.5
- Extract shared demo content to scripts/create-demo-content.sh, slim down
setup.sh and sandbox_playground_up.sh from ~150 duplicated lines each
- Move devcontainer scripts (ui, ui-p, docs) to .devcontainer/bin/
(source-controlled, survives make clean) with auto mode-switching
- Distinguish two dev workflows: make ui-dev (local Go) vs
dev-docker-up + pnpm run dev (no Go required)
- Add auto-init to dev profile so container doesn't crash on first run
- Add --cap-add ALL to playground volume init (cap_drop blocks chmod)
- Document port 19420 conflict between playground and dev profile
- Add docker-build vs docker-build-multiarch explanation and GHCR pull examples
- Increase git command timeout to 180s for constrained Docker networks
- Increase online test timeout to 300s
- Align Makefile help text formatting
Init now creates an initial commit (including .gitignore) so that
push/pull work immediately after setup. When a remote is configured,
init fetches and auto-pulls existing skills from the remote.
- Always create initial commit in doGitInit (fixes stale untracked
.gitignore blocking pull)
- Add ensureGitIdentity: set repo-local fallback when git user.name
is not configured, with hint to set global config
- Add tryPullAfterRemoteSetup: fetch remote after adding origin,
auto-pull if remote has content and local is empty
- Add hintGitRemoteError: context-aware hints for SSH, URL, and
network errors in push/pull
- Show Pro tip only when built-in skill is installed
- Sandbox: add --bare mode (skip auto-init), --volumes for down
(full reset), and sandbox-reset / sandbox:reset targets
- Docs: update init flow diagram, cross-machine-sync guide,
push/pull prerequisites, getting-started, troubleshooting
Update 5 website docs to cover the new --into install flag:
- commands/install.md: options table, quick examples, common scenarios
- guides/organizing-skills.md: new "Install Directly into Folders" section
- guides/project-setup.md: --into example in Step 3
- reference/file-structure.md: nested skill in source directory tree
- concepts/project-skills.md: --into in project directory structure
Reorganize playground demo skills into nested subdirectories
(security/, devops/, demos/, guides/) so users see the feature
out of the box.
Multiple agents (Amp, Codex, Copilot, Gemini, Goose, Kimi, OpenCode,
Replit) all map to .agents/skills in project mode. Previously init -p
showed each as a separate selectable entry pointing to the same directory,
which was redundant and confusing.
Now targets sharing a project path are collapsed into a single group entry
(e.g. "agents") with member names shown in parentheses. Discover uses
path-based dedup so an existing "amp" entry correctly prevents the
"agents" group from appearing as new. --select accepts both canonical
and member names for backward compatibility.
- Add project_name/project_path to agents meta-target in targets.yaml
- Add GroupedProjectTargets() in config package with unit tests
- Rewrite detect/list/prompt/discover in init_project.go
- Add agents target to playground demo project
Prevent test-docker and test-docker-online from running the same 36
offline integration tests. Each offline test file now has
`//go:build !online` so the two sets are mutually exclusive:
- default (no tag): 36 offline tests only
- -tags online: online tests only
Cover real GitHub operations that were previously untested:
- install clone (subdir skill)
- install --track (full repo)
- update tracked repo (freshly cloned → up to date)
- search with conditional GITHUB_TOKEN skip
Also pass GITHUB_TOKEN into sandbox-online container and remove
the -run filter so all online-tagged tests run automatically.
- new: redesign SKILL.md template based on Anthropic's skill-building
guide — description now includes WHAT+WHEN trigger format, step-based
instructions (Gather/Execute/Validate), examples, and troubleshooting
- search: prompt for keyword when run without arguments instead of
auto-browsing popular skills
- playground: default shell to $HOME (/sandbox-home) and mount source
as read-only to prevent accidental host file modifications
- Add (project) labels to status, sync, collect, target, and uninstall
headers for clear mode distinction (matching list/target list pattern)
- Add ui.WithModeLabel() helper for shared functions that run in both modes
- Update uninstall confirmation prompt: "from the project?" in project mode
- Add "Privacy-first" to README Why section and website homepage
- Replace "Unified Source" card with "Privacy-First" card on website
- Unify Quick Actions card heights with h-full in DashboardPage
- Add frontend rebuild note to Docker sandbox docs
Overhaul the documentation site with a hand-drawn design system:
- Custom CSS with paper textures, wobbly borders, and sketchy shadows
- Hero section with dashed ring, sparkle decorations, and floating animation
- Card components with post-it accent colors and tape decorations
- Replace README logo with hero-style version (ring + sparkles + paper bg)
Add "web dashboard", "skillshare ui" triggers to description for skill
activation. Add Web UI row to commands table. Create references/ui.md
with usage, flags, dashboard pages, project mode differences, REST API
endpoints, and Docker instructions.
The web dashboard now fully supports project-level skills. When launched
with `-p` or when `.skillshare/config.yaml` is auto-detected, the UI
adapts: Git Sync and Backup pages are hidden, Config edits the project
config, Install reconciles ProjectConfig.Skills[], and targets resolve
from project-level definitions.
Server: add NewProject constructor, IsProjectMode/configPath/saveConfig/
reloadConfig helpers, branch all handlers for project mode.
Frontend: AppContext propagates isProjectMode, Layout filters nav items,
Git/Backup pages show unavailable message, Config/Dashboard adapt.
Extract ReconcileProjectSkills to internal/config for reuse by both CLI
and server. Add unit tests and integration tests for project UI.
Update docs, README, and Docker playground with project mode demo project
and skillshare-ui-p alias.
- Add docker/sandbox/entrypoint.sh to auto-copy /ui-dist → internal/server/dist/
when missing, fixing the gitignore regression for all Docker workflows
- Add scripts/_sandbox_common.sh with shared SCRIPT_DIR, PROJECT_ROOT,
COMPOSE_FILE, and require_docker() function
- Refactor all 5 sandbox scripts to source _sandbox_common.sh
- Use YAML extension fields (x-sandbox-common) in docker-compose.sandbox.yml
to eliminate duplicated service config
- Break sandbox_playground_up.sh one-liner into readable steps, add
skillshare-ui alias and init hint
- Change test.sh to use `make build` for placeholder consistency
Single-binary web UI for visual skill management — React 19 + Vite
frontend embedded via go:embed, Go net/http REST API backend (30+
endpoints). Includes dashboard, skills browser, targets, sync, collect,
search, install, config editor, backup/restore, and git sync pages.
Also adds Bitbucket/GitLab URL parsing for install, git operations
library, skill version checking, CI frontend build steps, Docker
sandbox support, and website docs for the ui command.
Add complete integration test coverage for all CLI commands with
isolated sandbox environments that run on every change.
Changes:
- Add testutil package for sandbox test infrastructure
- Add 77 integration tests covering all commands (init, sync, status,
diff, backup, restore, pull, doctor, target, version, help)
- Update CI workflow to run integration tests on Ubuntu and macOS
- Add SKILLSHARE_CONFIG env var support for test isolation
- Add mise.toml for task management
- Move scripts to scripts/ directory