sujeet-pro
UserA plugin system for Claude Code that adds skills, agents, guidelines, repo-aware configurations, and MCP integrations to supercharge your development workflow.
Categories
Indexed Skills (16)
adk-explain
Explain, what-is, what's-the-difference-between, help-me-decide, I-don't-know, you-decide, what-would-you-recommend, not-sure, help-me-pick, teach-me. Advisor agent. Invoked directly OR hand-off target when any other skill detects user uncertainty. Job: teach the user how to choose without making the choice for them. Restates the question in plain English (defines jargon inline once), lays out 2–4 options with the consequence of each, surfaces the constraint that breaks the tie (the one question whose answer makes the choice deterministic), suggests a default ONLY if the user has ≥3 prior matches in their decision log (with quoted evidence), then waits. Never picks for the user. Never executes the underlying task — control returns to the calling skill after the user decides. Refuses out-of-scope topics (personal opinions, opinions about people). RAG-enriched when core.yaml.rag.enabled.
adk-implement
Implement, build, add, write, ship, code-up, wire-up code in an existing repo. Triggers on: Jira URL or KEY-NUM (specialized from-jira sub-flow — the most common path), GitHub issue URL or #N (from-issue), Confluence TDD URL (from-tdd), Slack thread permalink (from-slack-thread), or a freeform description (greenfield). Git mandatory; GitHub MCP optional but enables PR-by-URL flow. Every run: question-first (3 questions max), advisor-strategy plan with 2–4 trade-off options, edit-format discipline (SEARCH/REPLACE blocks per shared/edit-format.md), repo-native typecheck+lint+narrow-tests on each checkpoint. Writes plan/steps/diffs/report under `<repo>/.temp/adk/implement/<task>/`. Pushes only after explicit confirmation; never force-pushes; never merges; never touches a protected branch. Pulls Jira context via adk-mcp-atlassian, GitHub context via adk-mcp-github, and optional RAG via adk-mcp-rag. Supports --plan (read-only planning) → --act (writes). Sub-flows under references/.
adk-improve
Improve, learn, refresh-metadata, update-defaults, self-improve, train-skill, learn-from-session. The self-improvement loop. Always interactive (asks first: improve skill defaults from decision logs, metadata via MCP introspection, or both). For defaults: runs scripts/proposal_generator.py against `$ADK_DATA_HOME/improve/learning/decisions.jsonl`, drafts proposed updates to `$ADK_CONFIG_HOME/core.yaml.defaults.*`, presents each with ≥3 evidence lines, applies on confirm; rotates decisions.jsonl to learning/archive/ after each run; appends summary to learning/summary.md. For metadata: runs scripts/metadata_introspector.py to refresh `$ADK_DATA_HOME/improve/metadata/<source>.json` from every reachable MCP. Bounded: cannot change shared/constitution.md; cannot change Must-do/Must-not-do sections of any SKILL.md (those are constitution-grade). Each proposal requires per-item confirmation regardless of mode. Never auto-applies. Min-evidence default 3 (configurable). Manual-only by design (disable-model-invocation:
adk-pr-review
Deep PR review: tree-sitter AST chunking + ollama embeddings + LanceDB hybrid (vector + BM25) retrieval + SCIP cross-file symbols + harness-LLM reranker + feature-flow tracing + accept/reject/edit triage before posting. Triggers on a GitHub or Bitbucket Cloud pull-request URL OR no arg at all — when no URL is passed, the next eligible row from `$ADK_CONFIG_HOME/pr-queue.json5` is atomically claimed (FIFO by last_checked_at, 30-min auto-expiring `taken_at` lock so two terminals review different PRs). Curate the queue via `adk pr-scan` (scans Slack threads for PR links — main message AND replies — and upserts rows). When a URL is passed and that PR is already in the queue, the row's `slack` + `supporting_docs` are merged into the review context. **Global skill** — runs from anywhere; isolates to `$ADK_DATA_HOME/skill-pr-review/<repo>_pr-<n>/` (per `shared/paths.md`); never touches the cwd. Pipeline: clone+worktree at the PR head, tree-sitter chunker → ollama embed (`nomic-embed-text` default, `bge-m3` via `--de
adk-setup
Set-up, configure-overrides, init-config, refresh-metadata, verify-mcps, check-env. Stewards `$ADK_CONFIG_HOME/core.yaml` and the metadata cache. NOT a CLI-dep installer — brew, gh, jq, uv, node are the user's job (SETUP.md prints the exact commands). NOT a wiring tool — install.sh handles symlinks, MCP merges, hook wiring, AGENTS.md pointers. This skill picks up where install.sh stops: filling user data files (conversationally) and introspecting MCPs (with the agent's MCP client, which install.sh / curl cannot do). Four modes. --init: conversational scaffolding of core.yaml (workspaces, repos, data dictionary, RAG config). --enrich: queries every reachable MCP (Datadog dashboards, Statsig experiments, Mixpanel events, Snowflake schemas, Looker dashboards, Atlassian spaces, GitHub repos), summarizes findings, writes `enriched:` block + `$ADK_DATA_HOME/improve/metadata/<source>.json`. Never overwrites manually-set values. --check: superset of `scripts/adk_mcp_health.py` — also probes stdio MCPs (Atlassian via
adk-sync
Publish, sync, push-to, post-to, update, fetch-as-markdown, pull-from-Confluence/Jira/Slack. Bidirectional 3P bridge for markdown. Hybrid path: writes to `<repo>/.temp/adk/sync/<task>/synced/` when invoked from a repo with a repo-coupled doc, else `$ADK_DATA_HOME/sync/<task>/synced/` (default). READ mode (--read <url>): pulls a Confluence page / Jira description / GDoc / GitHub PR body / GitHub issue / Slack thread into local markdown. WRITE mode (--write <md-path> --to <destination>): publishes markdown to confluence / jira-desc / jira-comment / gh-pr-body / gh-issue-comment / slack / gdoc. Idempotent: match-by-id first, match-by-title-and-parent second, never by content hash. Format conversions are programmatic (md ↔ Confluence storage XHTML, md ↔ Jira ADF, md ↔ Slack blocks); AI only for "is this update safe?" checks. Per-invocation user confirmation required for every write regardless of mode (constitution §I.4). NEVER overwrites a human-authored target without explicit opt-in. NEVER changes sharing/restr
adk-document
Document, write, draft, write-up, summarize. Generates any professional markdown artifact: runbook, ADR, RCA, PR body, commit message, changelog, Mermaid diagram, README, migration guide, API reference, experiment report, incident summary, onboarding doc, design doc, handoff doc. Markdown-first; repo-bound: output goes to `<repo>/.temp/adk/document/<task>/draft.md` (or repo-canonical path with --write-to). Does NOT publish — that's /adk-sync. Composable: any other adk skill can call this to generate a writeup, then call /adk-sync to push it. Cites every non-trivial claim to a repo path or quoted source; no filler phrases (anti-pattern grep enforced). Audience-tuned (engineer / pm / exec / mixed) — voice doesn't mix. Loads adk-agent-doc-writer always; loads relevant shared/guidelines (observability for runbook/RCA, api-design for api-reference, security for RCA, accessibility for UI docs) by --type. Quotes from external sources capped at 15 words.
adk-investigate
Investigate, debug, why-is-X-slow/broken/down, what-changed, RCA, root-cause, post-mortem. Multi-source data investigator. Read-only. Triggers on: symptom + service ("checkout broken", "users see 500s" — incident sub-flow, most common), Datadog incident/monitor/dashboard/log URL (anchored), Slack alert permalink (auto-extracts service + symptom-time), Statsig URL (experiment or audit-log sub-flow), Mixpanel question (product-analytics), Snowflake/Looker question (data verification), `--use rca` for full root-cause analysis (combines incident + statsig audit-log ±2h + git-blame + optional mixpanel user-impact). Always pins an explicit time window — no "recent" / "lately". Correlates ≥2 independent signals before naming a root cause. States confidence on every claim (low/medium/high). Recommends lowest-blast-radius next action (rollback > flag-off > restart > investigate-which-PR > escalate). Never modifies a monitor / dashboard / gate / experiment. Refuses PII queries against columns in `connectors/<source>.md
adk-review
Review, audit, look-at, sanity-check, check any review-able target. Triggers on: GitHub PR URL (specialized review-pr — most common), `.` or local path (review-code-changes against the working tree), markdown file or Confluence URL (review-doc), comment-thread URL (review-comments), repo path with --audit (audit-repo), PR URL with --audit (audit-pr). Read-only by default. Produces severity-tiered findings (blocker / critical / should / may / nit) with `path:line` + ≤15-word evidence quotes from the actual file. Six dimensions in order: correctness → tests → security → performance → readability → consistency. Loads adk-agent-code-reviewer always, adk-agent-security-reviewer when diff touches auth/input/crypto/deps. Under --fix: applies accepted findings locally + pushes after confirm (never force, never merges, never protected branches). Under -i: walks each finding. Under --plan: read-only review-and-recommend; no edits. Refuses single-pass for diffs >5000 LOC.
adk-cli
Python package home for the `adk` CLI subcommands. NOT a slash-invokable skill — install.py skips this directory when symlinking skills into agent skill dirs. The `adk` binary at `<repo>/bin/adk` is symlinked to `~/.local/bin/adk` at install time and dispatches to the modules under `scripts/`.
document
Draft a technical document from an intent or a source. Triggers on "write the runbook / ADR / RCA / PR description / commit message / changelog / diagram / README / migration guide / API reference / experiment report / incident summary / onboarding doc / design doc". Markdown-first: produces a draft in a local file and NEVER publishes (no Confluence / Jira / Slack / GitHub posting — that is a separate concern). Reader-first voice: leads with the reader's question, cites every non-trivial claim to a repo path or quoted source, caps external quotes at 15 words, cuts filler. Audience-tuned (engineer / pm / exec / mixed) — the voice does not mix. GitHub context (PR / issue) is read via the gh CLI.
implement
Implement, build, ship, or wire up a change in an existing repo. Polymorphic on input: a Jira URL/KEY-123, a GitHub issue URL or #N (fetched via the gh CLI), a Confluence/Slack link (fetched via MCP), or freeform "build the X" prose. Plans before it acts: gathers context, presents 2-4 approaches with trade-offs, confirms, then writes the smallest correct change matching repo conventions, with tests for new behavior. Validates with the repo's own typecheck/lint/narrow tests at every checkpoint. Git via git directly (feature branch, never protected, never force, never merge); GitHub via the gh CLI only (PRs via gh pr create); clones SSH-only.
investigate
Investigate, diagnose, RCA, "why is X slow/broken/down", "what changed" for any production symptom. Triggers on a free-text symptom + service, a Datadog incident/monitor/dashboard/log URL, a Slack alert permalink, a Statsig gate/experiment URL, or a Mixpanel/Snowflake/Looker question. Read-only — never modifies a monitor, dashboard, flag, or experiment, and never triggers a rollback or restart (recommends; the human executes). Pins an explicit time window on every query, correlates >=2 independent signals before naming a root cause, and states confidence (low/med/high) with anchored evidence and <=15-word verbatim quotes. Produces a timeline + hypothesis + a lowest-blast-radius next action.
pr-review
HEAVYWEIGHT, GitHub-only deep PR review. Takes a GitHub PR URL, prepares a read-only git worktree at the PR head (SSH clone only), and reviews the diff with full cross-file context via Read/Grep/Glob. Fans out one agent per dimension (correctness, security, tests, performance, api, docs, observability, concurrency, feature-flow) through the Workflow tool, then adversarially verifies every finding before it survives. Traces feature flags / experiments via the adk-statsig MCP and pulls linked Jira/Confluence via the adk-atlassian MCP. Posts inline review comments, a review summary, and appreciations through the gh CLI after confirmation. NEVER merges, never force-pushes, never edits the PR's code. For a quick read-only pass with no worktree, use /adk:review.
review
Review, audit, sanity-check, look-at any review-able target. Triggers on a GitHub PR URL (fetched via the gh CLI), a local path or "." (review the working tree), a markdown/doc file, or a comment-thread URL. Read-only by default; --fix applies accepted findings locally and pushes to the PR branch after confirmation (never force, never merge, never to a protected branch). Produces severity-tiered findings (blocker / critical / should / may / nit) with path:line and <=15-word verbatim evidence quotes. Six dimensions: correctness, tests, security, performance, readability, consistency. For a deep PR review with cross-file code-context retrieval, use /adk:pr-review.
prj-docs
Generate and maintain this marketplace's Pagesmith documentation. Use when updating docs/, regenerating reference pages, adding Pagesmith navigation meta, polishing the home page, or documenting marketplace plugins, skills, agents, MCP servers, and helper binaries.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.