ccomkhj

Audit and improve AGENTS.md files in repositories, keeping CLAUDE.md as a one-line @AGENTS.md reference so the same project memory works across tools (Claude Code, Codex, Gemini CLI, …). Use when user asks to check, audit, update, improve, or fix AGENTS.md (or CLAUDE.md) files. Scans for all such files, evaluates quality against templates, outputs quality report, then makes targeted updates. Also use when the user mentions "AGENTS.md maintenance" or cross-tool "project memory optimization".

Graph a skill's flow — steps, artifacts, invocations — render it with draw-graphology, and report checkable weak links — each verified by two independent subagents — as a handoff file of lessons for improve-skills.

Summarize the CURRENT Claude session's work into a very concise 3P (Progress/Plans/Problems) Slack status report, draft it to the user for review, then place it as a native Slack draft for a final look before they send. User-invoked mid-session (`/slack-comm`) when you need to report what you just did in a tight, matter-of-fact format. Not for org-scale newsletters or pulling other people's updates.

Fold one lesson into one existing skill, behind a confirmation gate.

Unattended cleanup loop — scan the codebase for readability debt and run up to N targets (default 5) through a cleaner/reviewer/tester trio; verified cleanups land as per-target refactor commits on a clean-loop/<date> branch. Not a performance tool (pair-ratchet) and not for polishing just-written code (simplify-python).

Create data-driven network graph visualizations with graphology + sigma.js. Use when the user wants to visualize a network, dependency graph, knowledge graph, call graph, org/relationship map, communities/clusters, or any node-edge dataset too large to place by hand.

Use AFTER writing or editing Python to make it read better without changing behavior — idiomatic rewrites (comprehensions, enumerate/zip, f-strings, pathlib, dataclasses), flattened control flow (guard clauses, de-nesting, boolean logic), and removed cruft (dead code, redundant conditionals, pointless intermediates). High-confidence, behavior-preserving rules only; each carries an anti-rule for when NOT to apply it. Trigger when the user asks to "simplify", "clean up", "tidy", "de-nest", or make recently written Python more idiomatic/Pythonic. Not a performance tool (use pair-optimize) and not a behavior-changing refactor.

Fast in-session second opinion — an internal Claude subagent (default fable) reviews your proposal over N bounded rounds, then you decide. The single-engine, no-external-peer sibling of pair-consult.

Compact the WikiBrain memory vault back within its context budgets — merge, rewrite, and archive memories, applying only user-confirmed actions. Use when the user says "compact memory" or "clean up memory", or when more than a week has passed since the last `memory: compact` commit.

Recall past decisions and learnings from the WikiBrain memory vault — across all projects, its archive, and git history. Use when the user asks what they decided/learned/know about something, whether they've hit a problem before, or to search their memory.

Set up the WikiBrain memory vault on this machine — adopt project memory dirs via symlinks, install the wikibrain skills, ensure the private remote. Use after cloning the vault on a new machine, when a new project's memory should be adopted into the vault, or when the user says "set up wikibrain".

Sync the WikiBrain memory vault — commit and push to its private GitHub repo. Use when the user says "sync memory", after a session saves memories worth keeping, or when another wikibrain skill lands changes.

Use when the user asks about their portfolio — state, drift, rebalancing, P&L, adding/selling positions, editing targets or cash balance, swapping tickers, or troubleshooting CLI errors. Wraps the `portfolio` CLI of this repo with a structured assessment template and reference docs for every common task.

Use when the user wants a holistic, conversational review of their portfolio — "walk me through my portfolio", "how am I doing overall", "give me an overview and some advice", "what should I be thinking about", a periodic check-in, or a goal/risk gut-check. Runs a short guided Q&A (horizon, drawdown comfort, goal) every session, then synthesizes the live snapshot and income audit into a plain-language overview with prioritized, interpretive advice — and journals each review with a date so it can recall what you expected last time and flag what changed. Use this even when the user just says "review my portfolio" or "let's do a check-in" without naming metrics. NOT for one-off mutations (add/sell, edit targets), structured drift tables, or income/tax config — those belong to assess-portfolio and tune-income.

Use whenever the user wants to set up, enhance, or double-check the income estimate of this portfolio repo — the `income:`, `cash_interest_pct`, and `tax:` blocks in data/config.yaml that drive `portfolio income` (gross) and `portfolio income --net` (after-tax). Trigger on "estimate my dividends/interest", "my income shows 0 / n/a", "add a proxy", "set the cash interest rate", "show income net of tax", "set up Teilfreistellung", "is my income/tax config right?", "which distributing twin for <ETF>", or any request to verify/fix why an accumulating ETF reports no income or why the net figure looks off. Use this even when the user just says "check my income config" without naming the fields.

Use when implementing any feature or bugfix, before writing implementation code

Use when you have a spec or requirements for a multi-step task, before touching code

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP).

Use whenever the user asks to run, start, kick off, or trigger an agentune HPO campaign in natural language (e.g. "run HPO with lgbm 40 trials/round", "tune xgboost on housing", "optimize california_housing"). Enforces parameter elicitation, init, and the autonomous round loop.

Run bounded autonomous research loops in the cancer_detector repository. Use when Codex needs an autoresearch-style workflow for MRI segmentation or classification experiments by choosing one repo-native lane (`mri/cli/research.py`, `mri/cli/sweep.py`, `scripts/segmentation_autopilot.py`, `scripts/swinunetr_autopilot.py`, or a small hand-curated config wave), keeping a narrow editable surface, creating an untracked `experiments/autoresearch/tag-name/` workspace, launching dry-runs or SLURM jobs, reading manifests and reports, and deciding which changes to keep or discard.

Run the Continuous Improvement Harness interactively — clarify the goal for a target repo, then hand the TDD-gated improvement run off to a fresh Superset workspace.