Apply a security hardening baseline to a GitHub repository via gh api — branch and tag rulesets, Actions token defaults, secret scanning, Dependabot, private vulnerability reporting, CodeQL, immutable releases, and branch-restricted environments for deploy secrets. The write half of the pair with audit:security-audit…
Find and kill stale, leftover, or orphaned developer processes owned by the current user — abandoned dev servers (Vite / Vite+), editor LSP servers (oxlint), and duplicate or zombie MCP servers (playwright, context7, obsidian) left running after a shell, editor, or agent session exited.
Manage working documents in an Obsidian vault using a two-tier system — Inbox for quick dumps, Dev/Claude for persistent docs. Creates, finds, promotes, updates, and completes working documents with proper naming, headers, and lifecycle.
Fetch a GitHub issue by number or URL, analyze its scope of work, cross-reference local project docs and repo instruction files, check blocking relationships, and produce a structured implementation analysis with a task list.
Full GitHub issue lifecycle: pick an issue, create it, branch, commit, squash, push, PR, merge. Owns every git and gh write in the pipeline. Handles project board integration, base branch detection (main/master/epic-), and compact step reports. Supports entering at any step.
Draft or update a GitHub issue. Analyzes the description, asks clarifying questions, and produces a well-structured title and body following the project's template. Can update an existing issue or prepare a draft for filing later.
Produces a deep 11-section behavioral specification for a single module (typically 1–6 files, up to 1500 lines). Used by the code-to-spec pipeline on critical modules flagged by the scout, and as the primary agent for Small-tier bundles. Output is detailed enough that another engineer can reimplement the same…
Independent auditor for behavioral specs. Reads source files directly and evaluates how well a spec captures them — finds misses, errors, overreach, placeholders. Runs per-module and globally. Produces severity-grouped findings for the audit.md file.
Maps cross-module contracts — events, public exports, imports, integration points — across all modules in a bundle. Reconciles fire/listen sites, flags asymmetries, and produces the contract document used by the synthesizer. Use this agent after module analyzers complete, once per code-to-spec run.
Produces a medium-depth behavioral summary of a single module — purpose, public surface, state bindings, events, lifecycle, and key decisions. Use this agent once per module in the code-to-spec pipeline for Medium and Large tier bundles.
Surveys a source-code bundle (one directory up to an entire repo) and produces an architecture overview, module inventory, and a shortlist of critical modules for deep-dive analysis. Use this agent as the first pass of the code-to-spec pipeline on Medium or Large bundles.
Merges scout, per-module, contract-resolver, and deep-analyzer outputs into the final spec directory. Writes README.md, architecture.md, modules.md, contracts.md, and modules/ .md files. No source analysis — aggregation only. Use this agent at the end of the code-to-spec pipeline for Medium and Large tier bundles.
Attack a code change and report what survives. Dispatches independent reviewers in parallel, each with a single job and no access to the implementer's reasoning: one hunts correctness bugs blind to the issue, one hunts requirement gaps against the issue text, one runs outside the process entirely. Optionally…
Post-implementation cleanup of correct code. Prunes comments aggressively: deletes what restates the code, files design rationale to the commit message and wider context to the PR or an issue, collapses one mechanism explained in three places, strips issue and PR references that make a file unreadable on its own…
Extract a behavioral specification from a source-code bundle — a handful of files up to a whole 500+ file application. Produces reproducible, stack-neutral specs another engineer or LLM could use to rebuild the same observable behavior without reading the original source.
Approach board for a problem or a decision. Dispatches independent seats — some generating candidate approaches, including one agent outside this process entirely, some attacking the assembled set comparatively — then verifies the surviving objections and reports a ranked recommendation with its trade-offs. An…
Rule on a set of claims that already exists — review findings, a plan's assumptions, an analysis, a reviewer's objection — before anyone acts on them. Every claim gets exactly one verdict: it holds, it holds only in a narrower case, it is true but not worth the fix, it falls, or nobody can settle it. Whoever judges…
Handle a pull request from whichever side you are on. On someone else's branch it attacks the diff through changes-review, then publishes a real review with per-line comments and a verdict. On your own it works the threads people and bots left — decides its standing in each, verifies the claim before answering, runs…
Release an npm, pnpm, or bun package: version bump, validation, git tagging, and publishing, with safety checks and explicit user approval before anything is pushed.
Release workflow for the edloidas/skills collection. Validates git state, analyzes commits since the last tag, recommends a version bump (major/minor/patch), updates Claude and Codex packaging metadata, commits, tags, and pushes.