Skills

Scaffolds a conforming .claude/ artifact — rule, skill, hook, agent or output style — that already passes the repo's format checks, Anthropic's documented frontmatter shape, and the four token/context/determinism principles.

Add a book to the reading queue with Audible metadata + points-based ranking.

Monthly budget review with spending vs targets, trend analysis, commitment progress, and behavioral coaching.

Bi-weekly relationship check-in with Jessica — rotating frameworks (CBT/CFT/ACT-style), carryover from the previous session, open solo records logged between sessions, and a structured 60-min facilitation with agreements.

Ends the session — verify what changed, update the project, journal it, land every worktree this thread owns on its repo's trunk, report a line a script proved.

Cut a new public release of the Contextium template (github.com/Ashkaan/contextium) by editing the public repo directly to mirror this repo's current .claude/ AI layer, then committing + tagging + releasing.

Adversarial debate — three agents (Opus, Sol, Grok) argue a question from three positions, then the strongest conclusion is synthesized.

Turns a loose goal statement into a rigorous, codebase-wired `/goal` command (Claude Code's built-in agent-loop primitive) and fires it.

Build an automated HTML email — digest, report, alert, newsletter, notification, or any template sent through sendEmail / Resend.

EO Forum 5% preparation — interactive deepening to surface the most vulnerable share (top/bottom 5%) for this month's forum meeting.

Recommend restaurants from live Yelp data and bucket them into the food club.

Add a game to an HLTB list — Backlog (owned) or Purchase (not owned).

Books Amazon gift-card spend as real expenses in Monarch, from a paste of Amazon's Gift Card Activity page reconciled against Amazon's own closing balance.

Capture a moment of clarity, observation, or creative spark as a High Idea.

The "do" verb of the loop — execute a SPEC with rigorous self-validation.

Adversarial review of freshly-built code — find what was missed, what's inconsistent, what breaks.

Ingests an attached document of any kind — PDF, scan, photo, spreadsheet, Word file, saved email — by archiving the ORIGINAL to Google Drive and writing what it MEANS into knowledge/.

Chief of staff for work Ashkaan cannot be interrupted during — loads his decision charter, dispatches read-scoped specialist sub-agents, answers their questions ITSELF from the charter, and reports the outcome plus a log of the calls it made on his behalf.

Adversarial reasoning game that deals Ashkaan a legal-style fact pattern, takes his argument on both sides and his call, then steelmans the side he rejected and attacks until his reasoning holds or breaks.

Add a movie to Radarr so it downloads to Plex and enters the watch queue.

Create or update a person's row in the dashboard's people table, and their prose README in knowledge/people/, optionally enriching from Google Contacts.

Initializes a new WSP internal-app portal by cloning the closest-matching live portal (finance / operations / sales / dashboard) into `~/code/[new-name]`, stripping node_modules + .git, resetting name in package.json + wrangler.toml, and running `git init`.

Generates an on-demand world-forecast batch in-session from keyless live feeds, scored across four horizons and appended to a dated track-record log for later Brier grading.

Single entry point for project work — renders the live project index, or loads a slug and routes it to the think flow, /implement, /close or a state report, composing /spec, /implement and /close as primitives.

See the real running product of any app under ~/code, find its design problems, fix them, and explain what changed.

Runs the 6-phase Quarterly Review & Planning protocol — score the outgoing quarter's goals, archive a scored snapshot, and set the next quarter's SMART goals + Focus in knowledge/growth/goals/goals.md.

Reconcile Monarch Money accounts for a given month — balance-change vs transaction-sum gap detection, sync freshness, unreviewed-transaction count.

Weekly personal reflection — the week's time audit (every source, three-layer attribution, label learning), then a structured check-in with gratitude, open-concerns review, triage, five+ rounds of CBT/CFT/ACT inquiry, a clickable feeling pick, and an auto-close (journal + commit + push).

Restores Claude Code remote-control sessions killed by an OOM, a claude-remote.service restart, or a host reboot — back onto the phone, not just at-machine.

Writes one spec folder, specs/NNN-name/{spec,plan,tasks,research}.md, per designed roadmap row from spec-kit's templates, fills the row in ROADMAP.md, auto-dispatches spec-audit (reviewer consensus + spirit-check), folds findings, then auto-invokes /close.

Adversarial pre-review of any freshly-written or materially-rewritten SPEC or spec folder, running reviewer consensus (correctness) plus the spirit-check agent (fidelity to the user's verbatim ask — spec.md's Input and Clarifications, or a legacy SPEC's § 0 / § 0a — and to the behavior contract) in one pass and returning a converged SPEC plus a one-line commit-msg trailer.

Summarizes any video — meeting recording, YouTube, Loom, a local file — by reading what is ON SCREEN as well as what is said, so a diagram nobody narrates or a number only shown still lands in the summary.

Pursue a TCPA demand against a sender of unsolicited texts — archives the pasted text or attached screenshot as evidence, researches the legal contact, asks one batch of questions up front (the answers authorize the send), drafts an aggressive demand letter, and queues a live send from a@nulegal.com via the deployed tcpa-claim-sender task, tracking the claim as a knowledge/legal/tcpa/ record.

Files a TCPA regulatory complaint for an existing claim in knowledge/legal/tcpa/ — builds the submission from the claim's own records, traces every sworn assertion to its source before checking any perjury attestation, files the FCC informal complaint through the live form, and confirms the result from Ashkaan's inbox rather than the page.

Onboards or offboards a WSP team member across every code and repo surface a person exists in — the app tech rosters, the seven MCP access rosters, the ops portal, the rate card, the person card, the email signature, the stale-prone doc tables, and the KV indexes — then verifies each change against the system that owns the fact and hands the remaining console work to Emma (HR side) or Tyler (technical side) as a split checklist.

Works the alert approval queue one issue at a time, or solves a single pasted Uptime Kuma alert, failing schedule, or production outage.

Patches and reboots the whole homelab — fires a REAL infra-updater sweep (hosts, Docker stacks, Proxmox, TrueNAS, Home Assistant), follows it across the runner-killer handoffs, and reports what changed.

Checks every piece of content drafted on Ashkaan's behalf for a HUMAN READER — LinkedIn posts, blog articles, email, Slack messages, texts, client and team notes, job descriptions, outbound sequences, marketing copy — against the AI-voice standard it must pass.