Updated 1w ago
by highflame-ai
ai-factory
An opinionated, config-driven AI-SDLC toolkit for AI-assisted software engineering. Skills, agents, hooks, and a role-based capability model. The 1-100 toolkit for AI coding agents: spec-driven pipeline, adversarial review bench, deterministic gates.
21 GitHub stars (21)
| Period | New stars |
|---|---|
| Nov 21 | +6 |
| Nov 22 | +1 |
- 30 skills (30 skills found in its SKILL.md files)
- 4 forks (4 forks)
- 1w ago last update (Last updated 1w ago)
Install
Skills CLI
npx skills add highflame-ai/ai-factory
Asks which skills and agents to install. Just one: npx skills add highflame-ai/ai-factory --skill org-skill-skeleton — add -g for every project.
Claude Code
git clone https://github.com/highflame-ai/ai-factorycp -r "ai-factory/examples/org-skill-skeleton" ~/.claude/skills/
Copies the org-skill-skeleton skill into ~/.claude/skills/ for every project (use .claude/skills/ for one project). Repeat for the other skills listed below.
Plugin
/plugin marketplace add highflame-ai/ai-factory
Run inside Claude Code if the repository ships a plugin marketplace, then pick plugins from /plugin.
Skills can run code on your machine — read the SKILL.md first.
What it does
In the author's own words, from each skill's SKILL.md — showing the first 30 of 42 files.
-
org-skill-skeleton
Skeleton for an org-specific skill — copy into skills/, rename, and fill in. Encodes a recurring multi-step workflow your org performs (a deploy runbook, a release checklist, an onboarding scaffold).
examples/org-skill-skeleton/ on GitHub (opens in a new tab)SKILL.md -
add-detector
Scaffolds a new detection/analysis module in your platform's extensible service — module code, registry registration, config wiring, telemetry signal contract, and verification at every tier. Use when adding a new detection or analysis capability to a plugin-style service.
skills/add-detector/ on GitHub (opens in a new tab)SKILL.md -
adversary
Adversarial review of any artifact — a spec, architecture doc, task breakdown, diff/PR, plan, README, or plain prose claim. Assumes the artifact is wrong, broken, or incomplete, tries to prove it, and reports only the findings that survive its own refutation attempts. Read-only and artifact-agnostic. Use when the user says "attack this", "adversary", "red-team this spec/plan/PR", "try to break th…
skills/adversary/ on GitHub (opens in a new tab)SKILL.md -
analyze
Codebase health audit — identify technical debt, quality issues, and improvement opportunities
skills/analyze/ on GitHub (opens in a new tab)SKILL.md -
architect
Design architecture and break requirement into tasks
skills/architect/ on GitHub (opens in a new tab)SKILL.md -
audit-permissions
Review and clean up the standing Claude Code permission grants that accumulate in .claude/settings.json and settings.local.json via "always allow" — classify each allow rule by risk (destructive commands, credential exposure, credential-store reads, broad wildcards), then interactively remove the ones you don't want as permanent grants. Use periodically, or when `aif doctor`'s permissions-audit c…
skills/audit-permissions/ on GitHub (opens in a new tab)SKILL.md -
bugfix
End-to-end bug fix workflow — report, analyze, fix, verify, ship (PR + merge + deploy + knowledge capture)
skills/bugfix/ on GitHub (opens in a new tab)SKILL.md -
canary
Canary deployment with smoke tests — deploy to a zero-traffic revision, run health checks, and promote on success. Use when the user says "canary deploy", "deploy with canary", "smoke test the deploy", or wants deployment confidence before going live.
skills/canary/ on GitHub (opens in a new tab)SKILL.md -
debug
Systematic root-cause debugging for local repros, failing tests, and broken builds. Phase 1 = build a fast deterministic feedback loop (this is the skill); the rest is mechanical. Use when tests fail, builds break, or local behavior doesn't match expectations. For shared-environment issues use /triage-dev instead.
skills/debug/ on GitHub (opens in a new tab)SKILL.md -
dep-update
Safe dependency update loop — inventory outdated/vulnerable dependencies across the repo's toolchains, vet each candidate's changelog for breaking risk, update in an isolated branch, prove it with tests (and the org regression smoke when configured), and open one reviewable PR per coherent batch. Use for "update dependencies", "is anything vulnerable?", CVE bumps, or a scheduled hygiene pass.
skills/dep-update/ on GitHub (opens in a new tab)SKILL.md -
deprecate
Removes an old API, field, capability, or service across every repo declared in `.aif/config.yml` `repos:` in safe stages — announce → stop new uses → migrate existing → delete. Use when ripping out a legacy endpoint, removing a column, retiring a capability, or replacing X with Y. Pairs with `cross-repo-impact` to find every caller and `feature-prep` to retire the spec-registry entry.
skills/deprecate/ on GitHub (opens in a new tab)SKILL.md -
doc-drift
Find and fix documentation made stale by a code change — derive the changed public surfaces (APIs, CLI flags, config keys, env vars, endpoints) from a diff, hunt every doc that mentions them, classify stale vs missing coverage, and apply the doc updates. Use after a merge, before a release, or on "are the docs still right?".
skills/doc-drift/ on GitHub (opens in a new tab)SKILL.md -
expert
Create or update a domain expert — a curated, path-scoped context bundle distilled from the org's own docs and code, injected only when a change touches its domain. Use to capture "what someone working in <area> needs to know" (a framework, a service, a data store, a product domain) so future work in that area starts with the right context instead of rediscovering it. Distinct from references/ (s…
skills/expert/ on GitHub (opens in a new tab)SKILL.md -
feature-prep
At the start of any non-trivial feature, identify (1) whether it introduces or changes a platform capability/invariant that needs an entry in your org's spec registry (`org.spec_registry`) FIRST, and (2) where its tests belong (org regression suite vs repo-local). Invoke proactively before writing implementation code so the spec stays a leading indicator and the regression suite stays small + hig…
skills/feature-prep/ on GitHub (opens in a new tab)SKILL.md -
from-issue
Single entry point that takes a GitHub issue and drives it through to a merge-ready PR — classifies the issue, routes to /debug or /feature-prep (and /grill-feature when criteria fire), implements in vertical slices, adds tests in the right tier, runs /ship, opens the PR, and hands off to @pr-shepherd to babysit CI and review-bot comments. Bug fixes and small features both flow through. Use when…
skills/from-issue/ on GitHub (opens in a new tab)SKILL.md -
git-workflow
Git conventions — commit-subject prefixes that pass your CI's commit check (configured via `org.commit_prefix_regex`, defaulting to Conventional Commits), atomic commits, branch naming, when to worktree vs branch, and how to drive a PR through CI. Use when committing, branching, or about to open a PR. Pairs with the local `commit-prefix-check.sh` hook (which enforces) and the `pr-shepherd` agent…
skills/git-workflow/ on GitHub (opens in a new tab)SKILL.md -
grill-feature
Adversarial design review for architectural / cross-service / contract-touching work. Interviews the user one question at a time, anchored in whatever design sources exist (`.aif/context/architecture.md`, ADRs under `org.adr_dir`, the spec registry at `org.spec_registry`, per-repo CLAUDE.md), surfacing ambiguity in scope, contract, multi-tenancy, auth, observability, and migration order BEFORE im…
skills/grill-feature/ on GitHub (opens in a new tab)SKILL.md -
handoff
Compact the current session into a handoff document so a fresh agent (or you, after `/clear`) can pick up the work without losing context. Captures per-repo branch state across every repo touched (per `.aif/config.yml` `repos:`), in-flight PRs, MCP queries worth replaying, spec-registry entries in motion, and assumptions made. Use when context is getting heavy, before `/clear`, when stepping away…
skills/handoff/ on GitHub (opens in a new tab)SKILL.md -
incremental-implementation
Builds multi-file changes in thin vertical slices, one repo at a time, with a working state at every step. Use when implementing any feature that touches more than one file, crosses repo boundaries, or feels too big to land in one commit. Particularly important for multi-repo features where a partial change in one repo breaks contracts the next repo expects.
skills/incremental-implementation/ on GitHub (opens in a new tab)SKILL.md -
init
Bootstrap .aif/ structure in a new repo or subdirectory
skills/init/ on GitHub (opens in a new tab)SKILL.md -
license-audit
License and SBOM compliance audit — build a software bill of materials from the repo's actual dependency graph, resolve every component's license, judge each against the org's license policy (org.license_policy), and report allowed / denied / needs-review with evidence. Use on "are our licenses clean?", before a release or customer security review, or when a lockfile changes.
skills/license-audit/ on GitHub (opens in a new tab)SKILL.md -
manifest
Remote-derived view of all in-flight AIF work — open PRs and pushed feat/REQ-* branches across every session — with a coarse component/domain overlap report. Read-only and advisory.
skills/manifest/ on GitHub (opens in a new tab)SKILL.md -
measure
Measure whether the toolkit is actually paying off — compute delivery and quality metrics from git history, merged PRs, and .aif/ artifacts (PR cycle time, rework/revert rate, spec-to-merge time, review-finding trends, skill/telemetry usage), report them with honest caveats about attribution, and save a dated baseline to compare over time. Use to establish a baseline at adoption and to check impa…
skills/measure/ on GitHub (opens in a new tab)SKILL.md -
new-admin-module
Scaffolds a new CRUD module following your codebase's documented module pattern (schema → migration → repository → service → handler → module registration), wires it into the server, and publishes its interface to the shared resolver layer if other modules consume it. Use when adding a new feature area to your admin/API service.
skills/new-admin-module/ on GitHub (opens in a new tab)SKILL.md -
onboard
Adopt ai-factory in an existing codebase by analyzing it and DRAFTING the fill-in surfaces — the workspace service map, starter domain experts, conventions, and .aif/config.yml — for human review, instead of leaving a wall of blank templates. Use once per repo (or workspace) when first adopting the toolkit, after /init. Turns "fill in six files by hand" into "review and correct a draft."
skills/onboard/ on GitHub (opens in a new tab)SKILL.md -
optimize
API cost & performance scanner — identify expensive operations and optimization opportunities
skills/optimize/ on GitHub (opens in a new tab)SKILL.md -
proceed
End-to-end AIF pipeline that takes a requirement from spec through to deployed. Takes a REQ number as argument and runs validate → fix → architect → fix → implement → verify (reflect + review) → create PR → wrapup (merge, deploy, knowledge capture). Use when the user says "proceed", "proceed with REQ-xxx", "run the pipeline", "take REQ-xxx to completion", "implement REQ-xxx end to end", or wants…
skills/proceed/ on GitHub (opens in a new tab)SKILL.md -
reflect
Post-implementation self-review — catch issues before /review
skills/reflect/ on GitHub (opens in a new tab)SKILL.md -
release-notes
Draft multi-audience release notes (customer, engineering, executive) from a git tag or range — classifies commits and merged PRs by type, pulls context from linked specs, and produces a reviewable draft. Never publishes anything itself. Use when cutting a release, tagging a version, or when asked "what shipped since <tag>?".
skills/release-notes/ on GitHub (opens in a new tab)SKILL.md -
review
Multi-agent code review covering correctness, quality, architecture, test coverage, and security
skills/review/ on GitHub (opens in a new tab)SKILL.md
No skill matches that filter.
Review before you install
A skill is a set of instructions and scripts your agent will follow, and it can run code on your machine. Read the SKILL.md and any scripts first — this listing is not a security review.