Marketplace
Skills
Skills for Haarness: instructions and scripts an agent picks up when a task calls for them, added with one command.
Create PostHog Surveys
Designs, targets and launches PostHog surveys (in-app, hosted or headless) with defaults known to lift response rates.
PostHog Product Analytics
Instruments an app with PostHog event capture and user identification, installing the SDK first when it is missing.
PostHog Log Capture
Ships application logs to PostHog over OpenTelemetry, keeping existing log handlers and adding structured properties.
Create PostHog Experiments
Drafts a PostHog A/B test from the project's real events: hypothesis, flag, variant split, metric and running time.
Clean Up Stale Flags
Finds stale PostHog flags, checks they are safe to drop, removes the dead code paths and opens one draft PR per flag.
PostHog Feature Flags
Wraps new or changed code behind PostHog feature flags, with SDK patterns for web, server, mobile and plain REST.
PostHog Error Tracking
Turns on PostHog exception capture, adds manual captures at error boundaries and sets up source map uploads.
Explore LLM Traces
Debugs AI agent runs stored in PostHog: tool calls, context passed to the model, subagent steps, tokens and cost.
PostgreSQL Best Practices
Advises on Postgres schema design, indexing, query tuning, vacuum and replication, plus PlanetScale connections and CLI.
Vitess Sharding Guide
Explains Vitess keyspaces, VSchema, vindexes, online DDL and VReplication for sharded MySQL, especially on PlanetScale.
PlanetScale Neki
Covers running apps on Neki sharded Postgres: picking shard keys, avoiding fan-out queries, transactions and resharding.
MySQL Schema and Tuning
Makes careful MySQL changes: key and index design, EXPLAIN-driven query fixes, deadlock handling and safe online schema changes.
Commit Push and PR
Go from local changes to a draft pull request in one step: commit, push and fill in the PR description.
Audio Transcription
Get a text transcript of an audio or video recording, optionally with who said what, using OpenAI speech models.
Text to Speech
Generates narration, voiceover or IVR audio from text with OpenAI's speech models, singly or in batches.
Sentry Issue Lookup
Lists and inspects recent Sentry issues and events from the terminal with read-only queries, redacting personal data in output.
Repository Threat Modeling
Builds an evidence-based threat model of a repository covering trust boundaries, assets, abuse paths and mitigations.
Security Ownership Map
Mines git history to find who owns sensitive code, bus-factor hotspots and orphaned security files, with graph exports.
Notion Knowledge Capture
Save what a conversation decided or taught into the right kind of Notion page so the team can find it later.
Security Best Practices
Reviews Python, JavaScript/TypeScript and Go code against framework-specific secure-coding guides and writes a prioritized findings report.
Netlify Deploy
Links a project to Netlify and ships preview or production deploys through the Netlify CLI.
Desktop Screenshots
Captures the full screen, an app window or a region on macOS, Linux or Windows using bundled helper scripts.
Render Deploy
Analyzes a codebase, writes a render.yaml Blueprint or creates services via Render's MCP server, and checks the live deploy.
Playwright CLI
Automates a real browser from the shell with playwright-cli: snapshots, clicks, form fills, screenshots and traces.
Notion Spec to Tasks
Reads a Notion spec and creates a linked implementation plan and task pages, then keeps their status updated.
Notion Research Briefs
Searches across Notion workspaces and synthesizes findings into cited briefs, comparisons or reports.
Notion Meeting Prep
Pulls related Notion pages to draft meeting agendas and pre-reads tailored to attendees and decisions needed.
ASP.NET Core Guide
Guides building, reviewing and upgrading ASP.NET Core apps across Blazor, Razor Pages, MVC, Minimal APIs, SignalR and gRPC.
Test-Driven Development
Enforces a strict red-green-refactor loop: write a failing test, watch it fail, add the least code to pass, then clean up.
Jupyter Notebook Builder
Scaffolds clean experiment or tutorial notebooks from templates and tidies existing notebooks into a runnable top-to-bottom story.
Fix GitHub CI
Finds failing GitHub Actions checks on a pull request, pulls the logs, summarizes the cause and proposes a fix before changing code.
Address PR Comments
Collects review threads on the current branch's pull request, numbers them and applies fixes for the ones you pick.
Agent-Friendly CLI Builder
Designs and builds a durable command-line tool from API docs or scripts with JSON output, auth handling and a companion skill.
Writing Implementation Plans
Turns a spec into a step-by-step, test-first implementation plan with exact files, interfaces and verification commands per task.
Verification Before Completion
Makes the agent run the proving command and read its output before it claims tests pass, a bug is fixed or work is done.
Using Git Worktrees
Sets up an isolated git worktree for new work (after asking), installs project dependencies and confirms the tests start green.
MongoDB Query Optimizer
Explains why a MongoDB query is slow using explain output and Atlas Performance Advisor, then suggests indexes.
Systematic Debugging
Walks a bug through root-cause investigation, pattern comparison and a single tested hypothesis before any fix is written.
Subagent-Driven Development
Carries out a written plan by giving each task to a fresh implementer subagent, reviewing every task, then reviewing the whole branch.
Requesting Code Review
Sends a git commit range and its requirements to a fresh reviewer subagent and sorts what comes back by severity before moving on.
Receiving Code Review
Has the agent check each review comment against the real code, ask about unclear items and push back with reasons instead of agreeing.
Finishing a Branch
Once tests pass, offers to merge locally, push and open a pull request, or keep the branch, then tidies any worktree it created.
Executing Plans Inline
Works through a written implementation plan task by task in one session, test-first, keeping a progress ledger and ending with one review.
Dispatching Parallel Agents
Splits several unrelated failures or tasks into focused subagent jobs that run at once, then checks and merges their results.
Atlas Stream Processing
Builds, starts and debugs MongoDB Atlas Stream Processing pipelines fed by Kafka, change streams, S3 or HTTPS.
MongoDB Search and Vector
Plans Atlas Search, Vector Search and hybrid search indexes and queries, with an optional guided tour on sample data.
MongoDB Schema Design
Shapes MongoDB documents around how the app reads and writes data, deciding what to embed, reference, bucket or version.
KQL Query Expertise
Write Kusto queries for Azure Data Explorer, Log Analytics or Sentinel that run correctly and stay within limits.