Claude Skills by benthamite

Open the current Claude Code or Codex session log in Emacs via agent-log. Programmatic-only helper for explicit log-opening automation; not loaded in ordinary interactive sessions.

Audit agent configuration for redundancy, conflicts, dead weight, staleness, and instruction drift across CLAUDE.md or AGENTS.md, skills, memory, hooks, settings, and docs.

Discover, vet, and optionally install agent skills when the user asks "find a skill for X", "is there a skill that can...", "how do I add support for X", or otherwise wants reusable/installable agent capabilities. Do not use for ordinary task execution unless the user is explicitly exploring skills, tools, or capability extensions.

Triage and remediate GitGuardian secret incidents. Use for gitguardian/GG triage, secret audit, rotating leaked secrets, processing incidents, or clearing the GitGuardian dashboard.

Install and configure a new MCP server for Claude Code, including discovery, auth verification, credential-safe setup, scope placement, registration, sync, and restart/testing expectations.

Optimize project agent instruction files. Use for optimize/improve/clean up/review Claude.md or making persistent CLAUDE.md or AGENTS.md guidance more concise, concrete, and reliable; not for ordinary docs or code reviews.

Deploy and release a new version of the dotfiles Emacs profile. Use when the user says `/release-dotfiles X.Y.Z`, asks to deploy profile X.Y.Z, cut/publish/prepare a dotfiles release, or bump the dotfiles version. The agent builds and tests the unpinned development profile first, hands it to the user, and releases the pinned version only after the user confirms. Do not use for standalone Emacs package releases; use release-package instead.

Explicit workflow for auditing Claude/Codex skill behavior, implementation, or historical autonomous routing. Invoke only with /skill-audit in Claude or $skill-audit in Codex; never load automatically.

Resolve Claude/Codex skill files by name when neither the user nor the current session catalog identifies the artifact. Use for unresolved name-only lookup or resolver ambiguity; not when an exact path or a catalog root-alias mapping already identifies the requested skill.

Open the current Claude Code or Codex session log in Emacs via agent-log. Programmatic-only helper for explicit log-opening automation; not loaded in ordinary interactive sessions.

Audit agent configuration for redundancy, conflicts, dead weight, staleness, and instruction drift across CLAUDE.md or AGENTS.md, skills, memory, hooks, settings, and docs.

Discover, vet, and optionally install agent skills when the user asks "find a skill for X", "is there a skill that can...", "how do I add support for X", or otherwise wants reusable/installable agent capabilities. Do not use for ordinary task execution unless the user is explicitly exploring skills, tools, or capability extensions.

Triage and remediate GitGuardian secret incidents. Use for gitguardian/GG triage, secret audit, rotating leaked secrets, processing incidents, or clearing the GitGuardian dashboard.

Install and configure a new MCP server for Claude Code, including discovery, auth verification, credential-safe setup, scope placement, registration, sync, and restart/testing expectations.

Optimize project agent instruction files. Use for optimize/improve/clean up/review Claude.md or making persistent CLAUDE.md or AGENTS.md guidance more concise, concrete, and reliable; not for ordinary docs or code reviews.

Deploy and release a new version of the dotfiles Emacs profile. Use when the user says `/release-dotfiles X.Y.Z`, asks to deploy profile X.Y.Z, cut/publish/prepare a dotfiles release, or bump the dotfiles version. The agent builds and tests the unpinned development profile first, hands it to the user, and releases the pinned version only after the user confirms. Do not use for standalone Emacs package releases; use release-package instead.

Explicit workflow for auditing Claude/Codex skill behavior, implementation, or historical autonomous routing. Invoke only with /skill-audit in Claude or $skill-audit in Codex; never load automatically.

Resolve Claude/Codex skill files by name when neither the user nor the current session catalog identifies the artifact. Use for unresolved name-only lookup or resolver ambiguity; not when an exact path or a catalog root-alias mapping already identifies the requested skill.

Fetch recent tweets from a curated account list, triage them against its rubric, save an org digest, and update last-run state. Use twitter for direct lookups, twitter-discover for finding accounts, and twitter-vet for account scoring.

Discover high-value Twitter/X accounts in a niche through criteria, seeds, graph expansion, and vetting. Use for find accounts, who should I follow, curate a list, or twitter discovery; not for direct lookup, digests, or known-list vetting.

Vet, score, or decide whether to add Twitter/X accounts to a curated list. Use when asked to vet a handle, score accounts, decide whether an account belongs on a list, or when twitter-digest/twitter-discover need shared account scoring.

Fetch, read, summarize, or save specific Twitter/X tweets, threads, replies, profiles, timelines, follower lists, or searches. Use twitter-digest, twitter-discover, or twitter-vet for digests, account discovery, or account scoring.

Fetch recent tweets from a curated account list, triage them against its rubric, save an org digest, and update last-run state. Use twitter for direct lookups, twitter-discover for finding accounts, and twitter-vet for account scoring.

Discover high-value Twitter/X accounts in a niche through criteria, seeds, graph expansion, and vetting. Use for find accounts, who should I follow, curate a list, or twitter discovery; not for direct lookup, digests, or known-list vetting.

Vet, score, or decide whether to add Twitter/X accounts to a curated list. Use when asked to vet a handle, score accounts, decide whether an account belongs on a list, or when twitter-digest/twitter-discover need shared account scoring.

Fetch, read, summarize, or save specific Twitter/X tweets, threads, replies, profiles, timelines, follower lists, or searches. Use twitter-digest, twitter-discover, or twitter-vet for digests, account discovery, or account scoring.

Open the current Claude Code or Codex session log in Emacs via agent-log. Programmatic-only helper for explicit log-opening automation; not loaded in ordinary interactive sessions.

Evaluate HartreeWorks third-party skills one at a time. Use when the user says "/try-hartree-skills", asks to try Hartree skills, continue the Hartree skill queue, or record a verdict on a tested Hartree skill. Do not use for general skill discovery, local skill audits, or arbitrary third-party installs.

Process a list file strictly item by item, sequentially or with a bounded subagent pool. Use when the user needs no skipping, batching, or peeking ahead; not for lists that can be safely read all at once.

Use when adding works to Pablo's bibliography, creating BibTeX/BibLaTeX entries, adding DOI/ISBN/URL references, resolving missing citekeys, replicating zotra-extras-add-entry followed by ebib-extras-process-entry, or preparing notes that cite works not yet in the configured bibliography files.

Add or register one of the user's Emacs packages in my-emacs-packages.org, create its documentation note, and add its GitHub profile README card. Use when the user wants to update the public Emacs packages list for a specific package.

Design and build a reusable AI automation. Use for automate, create automation, turn this into a skill/command, new workflow, AI helper, or deciding whether a workflow should become a skill, directive, tool, or Emacs command.

Spec-based build workflow. Use for /build, build from spec, interview me about the spec, turn this rough spec into a plan, or developing spec.md into an implementation-ready plan before implementation.

Audit and revoke the sites Claude in Chrome is permanently allowed to access, across every Chrome profile. Use when the user asks to check/audit/review Claude's browser site permissions, asks which sites Claude can access, wants to revoke browser access to sensitive sites, or asks whether the "permanent but not sensitive" policy still holds. Not for Chrome's own content settings, extension host permissions, or OAuth app grants.

Audit code for actual defects, including bugs, application security issues, error handling gaps, and correctness problems. Not for readability, architecture, test coverage, performance-only, or broad environment security reviews.

Audit code architecture and maintainability for unnecessary complexity, premature or missing abstractions, duplication, unclear boundaries, and refactoring opportunities. Not for bug, security, or readability-only reviews.

Diagnose a specific agent, tooling, configuration, or workflow incident before any remedy. Use for /diagnose, why-did-this-happen questions, recurring process failures, broken hooks/skills/scripts, or diagnose-and-fix requests; not for ordinary code bugs, broad configuration audits, or repository-wide AI developer-experience reviews.

Audit, create, or refresh documentation for all Emacs extras packages in the dotfiles repo. Use when the user asks to document every extras package, fill missing extras docs, refresh outdated extras docs, check extras documentation coverage, or bring emacs/extras/doc up to date with emacs/extras sources.

Create or update an Org manual for an Emacs Lisp package in the Denote manual style. Use for documenting Elisp packages or emacs/extras; use generate-readme for Markdown README generation alone.

Route changes to emacs/config.org, dotfiles extras, Elpaca-managed package checkouts, and paired Claude/Codex configuration. Use to prepare external-package fixes, choose the canonical edit location, tangle the active Emacs profile, or run paired-config checks; not for read-only or unrelated dotfiles work.

Download the PDF of an academic paper (by DOI, publisher/doi.org URL, arXiv id, Anna's Archive md5, or title) with `paper-fetch`, the single paper-acquisition tool, including its browser fallback for challenge-protected hosts. Use whenever the user wants a paper, article, preprint or its PDF obtained, fetched, downloaded, saved or attached, even if they only paste a DOI or link, mention Anna's Archive, Sci-Hub, LibGen, PhilPapers or Unpaywall, or ask why a paper download failed. Not for books...

Audit a codebase's AI developer experience, including project instructions, verification commands, automation, onboarding traps, and collaboration friction. Not for ordinary bug, security, architecture, readability, or PR reviews.

Apply local coding and verification rules when modifying or testing Emacs Lisp, or when explicitly reviewing code against these conventions. Use for dotfiles extras, standalone Elpaca packages, and other .el files; not for non-Elisp work or read-only bug, security, or design reviews where these conventions are not material.

Use when the agent must execute or directly observe decisive live software acceptance through a real user-visible or runtime surface, including browser, Emacs, terminal display, async/network delivery, persistence, or scheduler behavior. Automation may drive the actual surface. Do not use solely to give manual test instructions; explain, write, or review; run an automated E2E suite; deploy or check deployment health/status; reproduce research results; or verify pure logic.

Use when Google Drive reports sync errors, a symlink or generated directory exists under ~/My Drive, or a Google Docs pointer fails to upload from the synced personal-account root.

Generate or update a GitHub-facing README.md for an Emacs Lisp package from its Org manual. Use when the user says "generate readme", "create readme", "update readme", "readme from manual", or wants a concise Markdown README for an Elisp package. Do not use for general project README writing or for packages without an Org manual; use document-elisp-package first when the manual is missing.

Handle Google Sheets comment review, replies, resolves, reopens, and notification emails correctly. Use this whenever the user asks to process, review, reply to, resolve, reopen, audit, verify, or summarize Google Sheets comments; mentions Google Sheets comment notification emails; asks to handle spreadsheet feedback; or opens/inspects a Google Sheet specifically to deal with comments. Do not use for ordinary spreadsheet editing or non-comment data analysis. This skill prevents treating Gmail...

Use when the user asks to save next steps, prepare a resume prompt, or continue work in a fresh session of the current tool, or when clear follow-up work should resume there.

Rewrite English prose to remove formulaic AI-style wording while preserving meaning, voice, quotations, citations and markup. Use for /humanize, de-slop, removing AI tells, "sounds like ChatGPT/Claude/Gemini", or final cleanup of AI-assisted prose. For writing under Pablo's name, use personalize, which incorporates this catalogue.

Audit code for intent transparency and reader comprehension. Use for understandable, readable, self-explanatory, naming, or future-maintainer clarity reviews; not for bug/security, architecture, DX, or lint-only reviews.