Provides web search and URL-to-text extraction using DuckDuckGo's keyless HTML endpoint, enabling web research without API keys.
Offers tools for inspecting local Git repositories, including status and commit history.
Includes a tool for viewing GitHub repository details, facilitating access to repository information on GitHub.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@sesha-skills Search my notes for 'meeting agenda' and show the results"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Canonical home of the
shesh_skillspackage and the Agent Skills library. An earlier banner here retired this repository in favour of shesh-core while it still shippedsrc/andtests/. Both repositories then published a package namedshesh_skillsand the console scriptshesh-skills-mcp, and the two copies drifted. The duplicate was removed from shesh-core; this repository is the single publisher.
Agent Skills library and everyday tool server for Shesh.
Licence: GPL-3.0-or-later
Layer: Mind (workflows) and Soma (actuators)
Part of: shesh-ecosystem
The skills below follow the Agent Skills specification, so they load unmodified in Claude Code, Codex, and other compliant agents.
skills/
├── audio-control/SKILL.md
├── audit-review/SKILL.md
├── autopilot/SKILL.md
├── backup-run/SKILL.md
├── bluetooth/SKILL.md
├── brightness/SKILL.md
├── calendar-check/SKILL.md
├── clipboard/SKILL.md
├── coding/SKILL.md
├── daily-briefing/SKILL.md
├── desktop-automation/SKILL.md
├── disk-cleanup/SKILL.md
├── display-control/SKILL.md
├── docs-writer/SKILL.md
├── documents/SKILL.md
├── file-organizer/SKILL.md
├── git-inspect/SKILL.md
├── gpu-mode/SKILL.md
├── messaging-send/SKILL.md
├── model-routing/SKILL.md
├── notes-capture/SKILL.md
├── notifications/SKILL.md
├── policy-inspect/SKILL.md
├── power-profile/SKILL.md
├── process-inspect/SKILL.md
├── safety-governance/SKILL.md
├── sandbox-run/SKILL.md
├── screen-recording/SKILL.md
├── screenshot-capture/SKILL.md
├── secrets-handling/SKILL.md
├── service-control/SKILL.md
├── session-control/SKILL.md
├── system-health/SKILL.md
├── system-updates/SKILL.md
├── thermal-check/SKILL.md
├── wallpaper/SKILL.md
├── web-research/SKILL.md
├── wifi/SKILL.md
├── window-appearance/SKILL.md
└── workspace-control/SKILL.mdSkill | Purpose |
| Control output volume, list audio devices, and switch the active sink |
| Inspect the audit log and verify its integrity |
| Make safe unattended progress on the backlog |
| Run, inspect, and prune backups |
| Pair, connect, disconnect, and list Bluetooth devices |
| Read and set screen brightness |
| Read the calendar and report what is scheduled |
| Read and replace the clipboard |
| Write, test, and refactor code safely |
| Produce the morning or evening digest |
| Read the screen and drive the desktop: accessibility tree, window targeting, screenshots, clicking, and typing |
| Reclaim disk space by clearing caches and reporting what is consuming storage |
| Inspect and change monitor resolution, refresh rate, scaling, and arrangement |
| Write or revise documentation in the house style |
| Read, convert, and extract text from PDF, Word, Excel, PowerPoint, and OpenDocument files |
| Sort files into folders by type, date, or project |
| Report repository state, recent history, and what has changed |
| Inspect and switch the hybrid graphics MUX between integrated and discrete mode |
| Send and read messages through connected bridges |
| Choose which model handles a task and inspect what is installed |
| Append to and search the Markdown notes vault |
| Send a desktop notification |
| Explain what the agent is currently permitted to do and why an action needs confirmation |
| Switch the system power profile between performance, balanced, and power-saver |
| Find what is consuming CPU, memory, disk, or network |
| The immutable safety layer governing destructive and irreversible actions. Always active. Requires confirmation before deleting data, force-pushing, modifying credentials, rewriting history, or acting outside the workspace. Overrides every other skill and cannot be refined away |
| Run an untrusted command inside a rootless container with no network |
| Start and stop screen recording |
| Take a screenshot of the screen, a window, or a region |
| Store and retrieve credentials without exposing them |
| Inspect and restart systemd units |
| Lock, suspend, hibernate, log out, reboot, or power off |
| Report overall machine state: load, memory, disk, temperature, and failed services |
| Check for and report pending system package updates |
| Report CPU and GPU temperature, fan behaviour, and thermal throttling |
| Set the desktop wallpaper |
| Research a topic from primary sources and report with citations |
| Inspect and change network connections, including Wi-Fi and airplane mode |
| Adjust window opacity, floating state, and fullscreen |
| Switch workspaces, move windows between them, and list what is open where |
safety-governance carries no allowed-tools field. Per the Agent Skills
specification that field grants permission and does not restrict: every tool
stays callable regardless of what it lists, so declaring a read-only set there
would describe a restriction that does not exist. Restriction comes from
disallowed-tools or the shesh-audit policy engine, which the skill's
compatibility field names. A test enforces the absence.
Skills are plain files with no Shesh-specific dependency. Copy a directory into any compliant agent's skill path:
cp -r skills/coding ~/.claude/skills/Skills follow progressive disclosure: listing costs only the name and description, and the body is read when a skill is selected.
from shesh_skills import skills
for s in skills.discover():
print(s.name, "-", s.description)
body = skills.get("coding").body()Resolution order, first match winning, so a user-installed skill overrides a shipped one:
$SHESH_SKILLS_DIR
$XDG_DATA_HOME/shesh/skills
The directory shipped with this package
A malformed skill is skipped rather than raising, so one bad directory cannot make the rest unavailable.
Related MCP server: Friday MCP Server
The stdio MCP server exposes the skill library plus everyday tools.
Tool | Purpose |
| Skill discovery and loading |
| Markdown notes vault |
| Search and retrieval, no API key |
| Repository inspection |
| Document conversion through pandoc |
| Desktop notification scheduler |
Every call passes through the audit guard, which returns allow, confirm, or deny and records the decision.
uv sync --extra dev
uv run pytest -q
uv run ruff check .
uv run shesh-skills-mcptests/test_skills_spec.py validates every skill against the specification:
directory layout, front-matter fields and limits, name and directory agreement,
body length, loader behaviour, and the read-only constraint on the safety skill.
Local-first. Search and retrieval use endpoints that need no API key.
Text only. No code ships with a skill, so a poor skill can degrade style but cannot act.
Least privilege. allowed-tools is scoped to what each skill needs.
Vulnerability reporting and the security posture are documented in the ecosystem security policy.
GPL-3.0-or-later — see LICENSE.
This server cannot be installed
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
Hosted MCP server for live public-data APIs and Skills for AI agents.
Personal assistant MCP server with search, execute, packages, jobs, secrets, and integrations.
Markdown-based note-taking with a hosted MCP server. Your notes serve you and your AI.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/gaganjainse/shesh-skills'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
