OutlineDriven

A polymorphic iterative repair loop that accepts verifier failures, structured findings, or bug descriptions, modifying and verifying until clean or an iteration cap is reached. Activated by commands like "fix" or "apply findings", it can auto-route to actions such as gh-fix-ci.

Mechanically tightens existing prose by restructuring sections based on dependency, splitting or merging paragraphs, and removing redundancy. This is used to compress verbose plan files, READMEs, ADRs, and design documents, without altering voice, register, tone, or any ODIN-mandated phrasing.

Personal taste skill employs 5 evidence-derived anchors ({anchor_names}) for prose, code, design, and decisions, operating in two modes: audit, which judges an artifact against its charter, and anchor, which loads the taste register before producing. It can be triggered with "{trigger_phrase}", "taste-test", "is this slop?", or "overkill?".

An evidence-first generator for a personal <name>-taste Claude Code skill. It mines local memories and agent histories for influences, slop bans, and overkill bans, asks compact confirmation forks, previews the synthesis, and then writes a right-sized taste skill with 5 anchors by default. Trigger with generate my taste skill, make my taste skill, or derive my taste spine.

Step up one layer of abstraction and surface a map of relevant modules and callers when the local view is too narrow. Trigger when the agent (or user) is unfamiliar with a code region and needs the surrounding architecture before committing to a change.

Inspect GitHub PR checks, retrieve failing GitHub Actions logs, summarize the failure, and then plan and implement a fix after user approval. This skill is used when debugging or fixing failing PR CI on GitHub Actions; external checks are reported as URLs only.

Addresses review/issue comments on the open GitHub PR for the current branch using the gh CLI. It first verifies gh authentication and prompts the user to log in if not authenticated.

Triage GitHub issues using a configurable label-based state machine. This is useful for triaging incoming issues, preparing them for an autonomous agent, or moving them between workflow states.

Analyze a codebase and create or improve an AGENTS.md file for future agent instances. Use when onboarding to a repository and capturing hard-to-rediscover conventions, constraints, and rationale.

Enforce idiomatic git-branchless during planning and executing tasks, utilizing detached-HEAD-first work, in-memory rebase via `git move`, event-log recovery via `git undo`, deferred branch creation, and speculative-merge `git sync` for base updates. This is useful for multi-commit work, history rewrites, stack edits, rebase/reorder, fixup insertion mid-stack, stacked-PR publishing, or recovering from mistakes.

Surface deepening refactors turn shallow modules into deep ones, informed by `CONTEXT.md` and `docs/adr/`. Use them when the user asks to improve architecture, find refactor candidates, raise testability, or make a codebase more agent-navigable, but skip for single localized fixes.

Produces share-safe copies of memory files under /tmp, redacting PII (paths, emails, session IDs, dates) and scanning for credentials (tokens, keys), without altering originals. Use when the user requests to sanitize memory for sharing, redact memory PII, or scan memory for credentials.

Decompose a task into independent concerns and execute them through broadly parallel, specialized agent groups. Use when a request involves multiple independent sub-tasks, research across separate domains, or work that can be parallelized across files or modules.

Software architect and planning specialist. Conducts thorough read-only planning, designing implementation plans, defining objectives, gathering relevant files, and summarizing tools before coding.

Automate Web-UI or human-triggered tasks by converting them into CLI and file-output loops, enabling LLMs to iterate independently with structured logs and scratchpads. Apply a trap-or-abandon approach: if a step isn't loopable, enhance the automation instead of manual oversight. Suitable for repetitive tasks, web UI triggers, or workflows with human-dependent inner loops.

Audits the memory directory for structural issues (orphans, duplicates, etc.) and staleness against session-history transcripts, reporting first and fixing on confirmation. Use when the user says "audit memory", "memory hygiene", or "find stale/duplicate memories".

Hotspot detection and performance root-cause analysis, including flamegraph interpretation, allocation tracking, latency profiling, and regression measurement. Use this skill when a workload misses its performance targets or before optimizing any critical path.

Scan agent's session-history transcripts for save-worthy signals (corrections, preferences, decisions, references), propose and write auto-memory files with valid frontmatter and MEMORY.md entry. Use when the user says "save this to memory", "remember that", or "scan this session for memories".

Merge multiple PRs into a temporal integration branch before merging to base, with ordered conflict resolution. Use when you want to validate a set of PRs together on a staging branch before advancing the base branch.

Review code changes on a given GitHub PR using gh CLI. Use when the user asks to review a pull request, analyze PR diffs, or provide feedback on open PRs with structured quality, security, and testing assessments.

Refactor by removing backward compatibility and legacy layers. Use when modernizing APIs, cleaning up migration debt, removing compat shims, or eliminating stale feature flags.

Merge one or more PRs into the base branch with queue-like sequencing and conflict resolution. Use when merging PRs that may conflict with each other or the base, requiring ordered application and intelligent conflict handling.

Bootstraps strict-mode tooling per ecosystem and provides per-task GOALS.md scaffolding for agentic loop self-verification. It configures typecheckers, linters, and schema validators for TS, Python, Rust, Go, and OCaml, establishing .agent.

Execute an implementation plan with surgical precision, typically after a planning phase has produced a step-by-step strategy and identified critical files. This skill focuses on precise code changes with verification at each step.

Proof-driven development: use when implementing with formal verification, utilizing property-based testing, theorem proving, or proof tactics, enforcing a zero unproven property policy.

Gathers read-only external knowledge using ODIN's 5-tier doc ladder (Official docs → API refs → Books/papers → Tutorials → Community). It's used for library APIs, framework behavior, SDK migrations, and verifies claims against primary sources.

Resolve code review comments by verifying validity, then proposing multiple architectural solutions (not naive fixes) for confirmed issues. Use when addressing review feedback or analyzing comment validity.

Installs Git pre-commit hooks via the project's hook tool (Husky+lint-staged, pre-commit, lefthook, cargo-husky) for commit-time formatting, linting, type-checking, or test gates, detecting the ecosystem first.

Initialize or idempotently revise the repo's .gitignore by composing gitignore.io templates, AI-tooling/IDE patterns, and confirmed noise from git status. Use when the user says set up gitignore, fix gitignore, or untracked files keep appearing in git status.

Reviews code changes on the current branch. This skill is invoked when the user requests a review of their current work, an analysis of recent commits, or a code quality assessment of the active branch against the main branch.

Adversarial security audit — STRIDE, OWASP Top 10, supply-chain (CVE/SBOM), secrets scan, auth/authz analysis. Use on changes touching auth, input parsing, deserialization, network I/O, dependencies, or secrets; before any production release or external-surface PR.

Reviews changed code across reuse, quality, and efficiency axes using parallel agents, then applies compress-op fixes. This is used for requests like 'simplify this diff/PR/branch' or 'compress these changes', distinct from general codebase cleanup or read-only reviews.

Build safe, syntax-aware srgn CLI commands for source-code search and transformation. Use for srgn commands, scoped refactors (comments/docstrings/imports/functions), multi-file rewrites with --glob, tree-sitter queries, or CI checks with --fail-any/--fail-none.

Test-Driven Development (TDD) across any supported language. Use when implementing features or fixes with TDD methodology, writing tests before code, or following XP-style development.

This cross-domain taste skill applies distinctive judgment to any artifact (prose, code, design, decisions), avoiding AI defaults. It features `audit` and `anchor` modes, auto-detected by phrasing or overridden with `/taste audit | anchor`, and triggers on queries like 'is this slop?' or 'elegant?'.

Write adversarial tests that intentionally stress failure paths. Use when hardening error handling, stress-testing assumptions, validating boundary behavior, or hunting silent failures.

Delete tests that don't catch real bugs, the inverse of TDD. A test is only justified if its removal would allow a real bug to reach production.

ODIN's compress-operations dispatcher under the Compressor/Extender role. Invoke on "tidy", "clean up", "tidy this file/memory/workspace/git/docs", or when active context (current file, diff, stack, memory directory) has structural rot to resolve before touching behavior. Detects target domain from context and routes to the sibling skill. Requires explicit target or clear active-context signal — d

Investigate a reported bug to root cause, then emit a TDD-shaped fix plan as an issue artifact. Trigger when the user reports a bug, says "triage", asks for issue investigation, or wants a fix plan before code changes.

Type-driven development. Use when developing with refined types, state machines encoded in types, or proof-carrying types; enforces totality and exhaustive pattern matching.

Validation-first development. Use when developing with formal state machine specifications, invariants, and temporal properties before writing implementation code.

Surface in-task-collaboration protocols when the user describes an AI workflow informally — URL-as-entity-reference, PR-comment threads as session memory. Trigger when the user names entities by colloquial label instead of stable URL, asks "how should I structure this for Claude", or describes a multi-step Claude workflow without a durable handle. Apply reactively, not as a checklist.

Verbalized Sampling (VS) protocol for intent exploration before planning, mode-aware. Default `exhaustive` runs full VS; `collaborative` runs tip-sharing dialogue; `adversarial` walks the design tree one fork at a time. Auto-detects from phrasing ("help me refine" → collaborative, "poke holes" → adversarial); override via `/askme adversarial|collaborative|exhaustive`. Use for ambiguous tasks or ma

Code search, analysis, and refactoring using ast-grep (sg). Use for AST-based code modifications, structural search, and linting.

Run the atomic-commit workflow on the current changes, then publish the resulting commits to the remote. Use whenever the user says \"commit and push\", \"ship these changes\", \"atomic commit and push\", \"publish my work\", or wants atomic commits delivered to origin in one step. Prefers `git submit` (git-branchless); falls back to a named branch + `git push origin HEAD:refs/heads/<branch>`. Ref

Design-by-Contract (DbC) development. Use when implementing with formal preconditions, postconditions, and invariants across any language.

ODIN's compact-form conversation skill uses a formal-logic English register with predicate claims, Hoare-triple framing, and ASCII shortened-English keywords. It triggers when the user requests "axiom", "axiom-mode", "axiom-compact", or "compact form".

An auto-router for context gathering that detects if a task requires codebase exploration or external knowledge research and dispatches accordingly. It triggers on requests like get context, background on X, context on X, how does X work, architectural orientation, or any setup-before-coding request.