rules
Skill257 starsChanged 18 days agoMITView on GitHub
Install and manage agent skills from Bitbucket repositories. Skills are directories containing a SKILL.md file that teach AI coding agents (Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, and many more) how to perform a task. This command group mirrors "gh skill" so the same workflow works for skills hosted on Bitbucket Cloud and Bitbucket Data Center. Repositories are addressed as WORKSPACE/REPO (Cloud) or PROJECT/REPO (Data Center) and accessed with the credentials of the active context. See https://agentskills.io/specification for the…
11 other projects use a version of this.
Adopted by
Most recently taken up by
- Beskars-Code/Skillstheir own version, first seen Nov 6, 2026
- Gabriel-Lafrance/Skillstheir own version, first seen Nov 6, 2026
- MagicJosh/rastheir own version, first seen Nov 2, 2026
- AcKeskin/contexturetheir own version, first seen Nov 1, 2026
- newkayak12/claude-skillstheir own version, first seen Nov 1, 2026
- wjgoarxiv/litopencodetheir own version, first seen Oct 13, 2026
Used with
Projects with it also use review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
SentrySkillreview
Fixture review skill.
Also in 4 projects that use this
Other versions11 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 11
- ByronFinn/dev-skills★ 4same name
- yinhunfeixue/software_development_skills★ 4same name
- newkayak12/claude-skills★ 3same name
- AcKeskin/contexture★ 2same name
- AltDoug/found-issues★ 2same name
- ibahgat/oh-my-iflow★ 2same name
- wjgoarxiv/litopencode★ 2same name
- MaGe1993/Claude-Code-Base-Harness★ 1same name
- MagicJosh/ras★ 1same name
- Beskars-Code/Skills★ 0same name
1 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 3 are distinctive, and 7 are very specific to this project.
0
Common
Most of the 10 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 105
Use --agent and --scope to control placement, or --dir for a custom directory.
- Very specific: line 128
When a skill name is given without a version, the newest tag is used, falling back to the default branch when the repository has no tags.
- Very specific: line 136
Use --all to install every discovered skill.
- Very specific: line 209
Use --agent to scan one host, --scope to scan only project or user scope, or --dir to scan a custom skills directory.
- Distinctive: line 270
When run with only a repository argument, the available skills are listed.
- Distinctive: line 340
no install metadata (metadata.bitbucket-*, metadata.github-*, local-path) is committed, since that records where a copy came from
- Very specific: line 346
Use --dry-run to validate without tagging, and --fix to strip install metadata from the committed files so you can review and commit the …
- Distinctive: line 397
Search for skill files across repositories in a Bitbucket Cloud workspace.
- Very specific: line 443
Check installed skills for updates by comparing the commit recorded in each SKILL.md frontmatter (metadata.bitbucket-commit) against the …
- Very specific: line 453
Use --unpin to clear the pin and include them in the update.
“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<!-- auto-generated by cmd/docgen — do not edit -->bkt skill
Install and manage agent skills from Bitbucket repositories.
Skills are directories containing a SKILL.md file that teach AI coding agents (Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, and many more) how to perform a task. This command group mirrors "gh skill" so the same workflow works for skills hosted on Bitbucket Cloud and Bitbucket Data Center. Repositories are addressed as WORKSPACE/REPO (Cloud) or PROJECT/REPO (Data Center) and accessed with the credentials of the active context.
See https://agentskills.io/specification for the skill format.
bkt skill <command> [flags]Examples
# Install a skill from a Bitbucket repository
bkt skill install myteam/agent-skills code-review
# List installed skills
bkt skill list
# Preview a skill before installing
bkt skill preview myteam/agent-skills code-review
# Update all installed skills
bkt skill update --allSubcommands
| Subcommand | Description | Key Flags |
| install | Install agent skills from a Bitbucket repository | --agent, --all, --allow-hidden-dirs, --dir |
| list | List installed skills | --agent, --dir, --scope |
| preview | Preview a skill from a Bitbucket repository | --allow-hidden-dirs |
| publish | Validate skills and tag a release | --dry-run, --fix, --message, --tag |
| search | Search for skills in Bitbucket Cloud code | --limit, --workspace |
| update | Update installed skills to their latest versions | --all, --dir, --dry-run, --force |
bkt skill install
Install agent skills from a Bitbucket repository or local directory into your local environment. Skills are placed in a host-specific directory at either project scope (inside the current git repository) or user scope (in your home directory, available everywhere).
Supported --agent values:
- GitHub Copilot (github-copilot)
- Claude Code (claude-code)
- Cursor (cursor)
- Codex (codex)
- Gemini CLI (gemini-cli)
- Antigravity (antigravity)
- Antigravity CLI (antigravity-cli)
- Antigravity 2.0 (antigravity2.0)
- AdaL (adal)
- Amp (amp)
- Augment (augment)
- IBM Bob (bob)
- Cline (cline)
- CodeBuddy (codebuddy)
- Command Code (command-code)
- Continue (continue)
- Cortex Code (cortex)
- Crush (crush)
- Deep Agents (deepagents)
- Devin (devin)
- Droid (droid)
- Firebender (firebender)
- Goose (goose)
- Grok (grok)
- iFlow CLI (iflow-cli)
- Junie (junie)
- Kilo Code (kilo)
- Kimi Code CLI (kimi-cli)
- Kiro CLI (kiro-cli)
- Kode (kode)
- MCPJam (mcpjam)
- Mistral Vibe (mistral-vibe)
- Mux (mux)
- Neovate (neovate)
- OpenClaw (openclaw)
- OpenCode (opencode)
- OpenHands (openhands)
- Pi (pi)
- Pochi (pochi)
- Qoder (qoder)
- Qwen Code (qwen-code)
- Replit (replit)
- Roo Code (roo)
- Trae (trae)
- Trae CN (trae-cn)
- Universal (universal)
- Warp (warp)
- Zencoder (zencoder)
Use --agent and --scope to control placement, or --dir for a custom directory. The default scope is "project" and the default agent is "universal", which installs into the shared .agents/skills directory read by GitHub Copilot, Cursor, Codex, Gemini CLI, Antigravity, Amp, Cline, OpenCode, Warp and others. Agents that resolve to the same destination share one copy.
The first argument is a Bitbucket repository: WORKSPACE/REPO on Bitbucket Cloud, PROJECT/REPO on Bitbucket Data Center, or a full repository URL. The active context (or --context) supplies the host and credentials. Use --from-local to install from a local directory instead; files are copied (not symlinked) with local-path tracking metadata injected into frontmatter.
Skills are discovered using the skills//SKILL.md convention from the Agent Skills specification (https://agentskills.io/specification), including namespaced skills/{author}//SKILL.md, plugins//skills//SKILL.md, root-level */SKILL.md, and a skills/ directory nested under a prefix.
The skill argument can be a name, a namespaced name (author/skill), or an exact path within the repository (skills/author/skill, packages/agent-skills/code-review, or any .../SKILL.md path). Passing an exact path skips the full repository listing, which is faster on large repositories.
When a skill name is given without a version, the newest tag is used, falling back to the default branch when the repository has no tags. To pin a version, append @VERSION to the skill name or use --pin; the version is resolved as a tag, branch, or commit SHA.
Installed skills carry source metadata (metadata.bitbucket-*) in their SKILL.md frontmatter so "bkt skill update" can detect changes.
Use --all to install every discovered skill. Without a skill name or --all, the available skills are listed (tab-separated when piped) so you can browse or grep them before re-running with a specific skill.
Alias: add
Usage
bkt skill install <repository> [<skill[@version]>] [flags]Flags
| Flag | Short | Description |
--agent | Target agent (see supported values above) | |
--all | Install all skills in the repository | |
--allow-hidden-dirs | Include skills in hidden directories (e.g. .claude/skills/, .agents/skills/) | |
--dir | Install to a custom directory (overrides --agent and --scope) | |
--force | -f | Overwrite existing skills without prompting |
--from-local | Treat the argument as a local directory path instead of a repository | |
--pin | Pin to a specific tag, branch, or commit SHA | |
--scope | Installation scope: project or user | |
--upstream | Install from the upstream source when a re-published skill is detected |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# List the skills available in a repository
bkt skill install myteam/agent-skills
# Install a specific skill into .agents/skills of the current repository
bkt skill install myteam/agent-skills code-review
# Install all skills from a repository
bkt skill install myteam/agent-skills --all
# Install a specific version
bkt skill install myteam/agent-skills code-review@v1.2.0
# Install by exact path (skips the full repository listing)
bkt skill install myteam/agent-skills skills/monalisa/code-review
# Install for Claude Code at user scope
bkt skill install myteam/agent-skills code-review --agent claude-code --scope user
# Install from a Data Center repository by project key
bkt skill install PROJ/agent-skills code-review
# Install from a local directory
bkt skill install ./my-skills-repo code-review --from-local
# Include skills kept in hidden directories such as .claude/skills/
bkt skill install myteam/agent-skills --allow-hidden-dirsbkt skill list
List installed agent skills across known agent host directories.
By default, scans all supported agent hosts in both project and user scope. Use --agent to scan one host, --scope to scan only project or user scope, or --dir to scan a custom skills directory.
Project-scope skills are discovered relative to the current git repository root. User-scope skills are discovered relative to your home directory. Skills authored in the current repository's skills/ directory are listed as published. Skills installed by other tools (for example gh) are listed with the source they record.
Alias: ls
Usage
bkt skill list [flags]Flags
| Flag | Short | Description |
--agent | Filter by target agent (see bkt skill install --help for values) | |
--dir | Scan a custom directory for installed skills | |
--scope | Filter by installation scope: project or user |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# List all installed skills
bkt skill list
# List skills installed for Claude Code
bkt skill list --agent claude-code
# List user-scope skills
bkt skill list --scope user
# List skills as JSON
bkt skill list --jsonbkt skill preview
Show a skill's files and SKILL.md content in the terminal without installing anything. The output goes through the configured pager when stdout is a terminal.
A file tree is shown first, followed by SKILL.md and the skill's other files (up to 20 files or 512 KiB).
When run with only a repository argument, the available skills are listed.
The skill argument can be a name, a namespaced name (author/skill), or an exact path within the repository (skills/author/skill, packages/agent-skills/code-review, or any .../SKILL.md path).
To preview a specific version, append @VERSION to the skill name. The version is resolved as a branch, tag, or commit SHA.
Alias: show
Usage
bkt skill preview <repository> [<skill[@version]>] [flags]Flags
| Flag | Short | Description |
--allow-hidden-dirs | Include skills in hidden directories (e.g. .claude/skills/, .agents/skills/) |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# Preview a specific skill
bkt skill preview myteam/agent-skills code-review
# Preview a skill at a specific version
bkt skill preview myteam/agent-skills code-review@v1.2.0
# Preview exactly what an install wrote, by commit SHA
bkt skill preview myteam/agent-skills code-review@abc123def456
# Preview from a non-standard nested path
bkt skill preview myteam/skills-repo packages/agent-skills/code-review
# List the skills available in a repository
bkt skill preview myteam/agent-skillsbkt skill publish
Validate a repository's skills against the Agent Skills specification and, with --tag, mark the current commit as a released version.
Skills are discovered with the same conventions as install: skills//SKILL.md, skills/{author}//SKILL.md, plugins//skills//SKILL.md, root-level */SKILL.md, and a skills/ directory nested under a prefix. Skills in hidden directories are ignored: those are copies installed by an agent, not the repository's own work.
Validation checks:
- the skill name follows the agentskills.io naming rules
- the skill name matches its directory name
- the required frontmatter fields (name, description) are present
- allowed-tools is a space-delimited string, not a list
- no install metadata (metadata.bitbucket-, metadata.github-, local-path) is committed, since that records where a copy came from
Recommendations are reported as warnings: a license field, a description under 1024 characters, and a body under 500 lines.
Use --dry-run to validate without tagging, and --fix to strip install metadata from the committed files so you can review and commit the result.
Bitbucket has no releases, so publishing a version means creating a tag on the current commit. "bkt skill install" resolves an unversioned request to the newest tag, so tagging is what makes a version installable.
Usage
bkt skill publish [<directory>] [flags]Flags
| Flag | Short | Description |
--dry-run | Validate without tagging | |
--fix | Strip committed install metadata without tagging | |
--message | Annotation for the tag (defaults to the tag name) | |
--tag | Tag the current commit with this version (e.g. v1.2.0) |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# Validate the skills in the current repository
bkt skill publish --dry-run
# Strip committed install metadata, then review the changes
bkt skill publish --fix
# Validate and tag the current commit as v1.2.0
bkt skill publish --tag v1.2.0
# Validate a repository checked out elsewhere
bkt skill publish ./agent-skills --dry-runbkt skill search
Search for skill files across repositories in a Bitbucket Cloud workspace.
The query uses Bitbucket Cloud code-search syntax. Scope it further with terms such as repo:agent-skills or path:skills. The workspace comes from --workspace or the active context. Results are restricted to SKILL.md files. Bitbucket Data Center is not supported because it has no public workspace code-search API.
Note: Atlassian has announced that the Cloud code-search REST endpoint will be deprecated on November 1, 2026.
Usage
bkt skill search <query> [flags]Flags
| Flag | Short | Description |
--limit | -L | Maximum matches to return |
--workspace | Bitbucket Cloud workspace (defaults to context workspace) |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# Search the active context's workspace
bkt skill search "code review"
# Search one repository and emit structured output
bkt skill search "review repo:agent-skills" --workspace myteam --jsonbkt skill update
Check installed skills for updates by comparing the commit recorded in each SKILL.md frontmatter (metadata.bitbucket-commit) against the latest commit that touched the skill in its source repository, and reinstall the ones that changed.
Scans every known agent host directory in both project and user scope, or a single directory with --dir. Without arguments, checks all installed skills; with skill names, checks only those.
Pinned skills (installed with --pin or @version) are skipped with a notice. Use --unpin to clear the pin and include them in the update.
Skills without Bitbucket metadata (installed manually or by another tool) are prompted for their source repository in interactive mode. With --all or in non-interactive mode they are skipped with a notice. Skills installed by gh are listed with a reminder to update them with gh.
With --force, skills are re-downloaded even when already up to date. This overwrites locally modified files with their original content and removes extra files added locally.
In interactive mode, the available updates are shown and confirmed before proceeding. With --all, updates are applied without prompting. With --dry-run, updates are reported without modifying any files.
Usage
bkt skill update [<skill>...] [flags]Flags
| Flag | Short | Description |
--all | Update all skills without prompting | |
--dir | Scan a custom directory for installed skills | |
--dry-run | Report available updates without modifying files | |
--force | Re-download even if already up to date | |
--unpin | Clear pinned version and include pinned skills in update |
Inherited Flags
| Flag | Short | Description |
--context | -c | Active Bitbucket context name |
--format | Output format: json or yaml (alias for --json/--yaml) | |
--jq | Apply a jq expression to JSON output (requires --json or --format json) | |
--json | Output in JSON format when supported | |
--template | Render output using Go templates | |
--yaml | Output in YAML format when supported |
Examples
# Check and update all skills interactively
bkt skill update
# Update specific skills
bkt skill update code-review git-commit
# Update all without prompting
bkt skill update --all
# Re-download all skills (restore locally modified files)
bkt skill update --force --all
# Check for updates without applying
bkt skill update --dry-run
# Unpin skills and update them to latest
bkt skill update --unpinDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

