Product & Planning

mahdtech Bundle

Triage a GitHub backlog - move issues and external PRs through a state machine of triage roles (categorise, verify, grill if needed, write agent-ready briefs) using the gh CLI and GitHub labels. Use when triaging your GitHub issues and PRs, deciding what's ready for an agent or a human, or turning a backlog into agent-ready briefs.

mahdtech Skill

Turn the current conversation into a spec (sometimes called a PRD) and publish it to your project's issue tracker - no interview, just synthesis of what you have already discussed.

mahdtech Bundle

Review the changes since a fixed point (a commit, branch, tag, or merge-base) along two axes - Standards, meaning does the code follow this repo's documented conventions plus a baseline of common code smells, and Spec, meaning does the code do what the originating issue, ticket, or PRD asked for. Runs both reviews as parallel sub-agents and reports them side by side without merging or reranking them. Use when the user wants to review a branch, a pull request, work-in-progress changes, or asks to review the diff since some point.

mahdtech Bundle

Run the TARS software factory unattended over one repository's backlog. The agent becomes the foreman: it drives the Antigravity CLI (agy) headlessly through batch runs, peer reviews, and rework until the backlog drains or a human is needed.

mahdtech Skill

Groom and prioritise a backlog - order, cut, cluster, and sequence a pile of work by value against cost and risk so the next thing to do is obvious. Use when the user wants to prioritise or rank a backlog, decide what to build next, trim or triage a pile of ideas, apply a lens like value-vs-effort, RICE, or cost-of-delay/WSJF, or sequence work around dependencies.

mahdtech Skill

Coordinate the full backlog lifecycle by sequentially executing tars-backlog-audit, tars-backlog-triage, and tars-backlog-implement to resolve all issues. Reach for this when asked to run a full backlog loop, converge on a complete project goal, or manage the overall ticket pipeline.

mahdtech Skill

Review the code implementation of a backlog ticket on a subagent branch, assessing compliance with spec/acceptance criteria and repo standards before merge. Reach for this during the implementation phase of the backlog loop.

mahdtech Skill

Audit the codebase for bugs, features, security issues, or technical debt, and generate structured issue files in `.tars/issues/todo/`. Reach for this when requested to perform a codebase audit, search for bugs and tasks, or populate the backlog.

mahdtech Skill

Triage pending backlog issues in `.tars/issues/todo/` to verify their accuracy, check for hallucinations, and add review notes. Reach for this when requested to triage tickets, verify backlog accuracy, or prepare issues for implementation.

mahdtech Skill

Prepare to run the tars-backlog-loop by verifying repository integrity, resolving isolated spoke workspaces, and cleaning up orphaned clones and branches. Reach for this to reset the environment before starting a full backlog loop.

mahdtech Bundle

Implement pending backlog issues from `.tars/issues/todo/` in parallel, conflict-free batches using isolated clones. Reach for this when asked to implement backlog issues, execute tasks from tickets in parallel, or resolve the issue queue.

mahdtech Skill

Use when creating a new backlog issue/ticket in the `.tars/issues/todo/` directory, defining its YAML frontmatter, headings, tasks, acceptance criteria, evidence collection, and triage review expectations.

mike007jd Skill

Use when turning game audit findings or known issues into a prioritized repair plan.

bitranox Bundle

Use at the end of a turn that produced a learning, such as a correction from the user, a rule or preference stated ("remember this", "from now on", "always/never"), a process or tooling mistake (wrong command, a quirk of the shell/SSH/environment, a misread of stale output, over-waiting, a tool or file you missed and re-derived), a wasted build or test cycle, or any reusable discovery (a procedure, a timing, a gotcha, a flag combination, a path). A gated Stop hook nudges this automatically; also run on "self-improve", "/self-improve", "improve the harness", or "capture what we learned".

bitranox Bundle

Use when a session's context is large enough that quality is degrading or compaction is close, when a Stop nudge reports the handover threshold was crossed, or on "write a handover", "hand this over", "context is getting full", or "let's start a fresh session". Also use when unfinished work or an unmet user request must outlive the session and belongs in the standing backlog OPEN-WORK.md, when open items keep sinking under whatever was worked on last, or on "add this to the backlog", "what is still open"

fatih-developer Bundle

Orchestrate a B2B founder from market uncertainty to validated demand, first paying customers, repeatable revenue, retention, and growth. Use when a user wants to validate, position, market, sell, price, retain, or grow a B2B SaaS/product; when they ask what to do next in go-to-market; or when continuing an existing B2B marketing roadmap. Diagnose the current stage, find the single biggest bottleneck, execute the work tools permit, produce stage artifacts, validate evidence, and only then advance.

fatih-developer Bundle

Design the first-run experience (FRE) for mobile apps. Sequence permission requests, define feature discovery flows, plan empty states, and create retention-focused day 1-3 scenarios.

fatih-developer Bundle

Force establishing and preserving the primary focus category before discussing any new project, product, web app, AI tool, business initiative, workflow, or job. Use when the user starts building, planning, evaluating, improving, or substantially changing something and the primary focus is not explicit. Do not proceed to features, tools, architecture, strategy, roadmap, or solutions until the user chooses one primary focus. Especially useful for non-technical users.

forgeyclap Bundle

Prompt Master intake — captures the real goal via one clarifying-question list before building. Use at the START of any /forge build/create/automate task — intake, requirements, scope, PRD.

forgeyclap Skill

Read-only full-codebase priming scanner. Use before a PRD or big refactor, or onboarding into an unfamiliar project — deep learn, prime the codebase, scan the project, risk list.

forgeyclap Skill

Generate a structured PRD (markdown + meta) for a non-trivial build and auto-create a ticket per acceptance criterion (Forge Mission Control Phase 2, WP3 "PRD generator"). Use before starting a non-trivial feature/build, typically right after Deep Learn Mode. Trigger on "write a PRD", "generate a PRD", "scope this feature", "spec this out before we build".

afine907 Bundle

Large-scale PRD project orchestration engine with risk-first audit and DAG task execution. Trigger: user says "/task-loom init"、"大型项目编排"、"PRD 分解".

asdc163 Bundle

Turn a verified product or skill release into an evidence-backed, channel-aware English share pack with a clear user job, proof ledger, try path, boundaries, feedback ask, and learning writeback. Use when a maintainer needs public discovery material without inventing adoption or growth claims.

asdc163 Bundle

Turn multiple evidence-backed opportunity candidates into one bounded product bet with a source ledger, user job, assumptions, opportunity cost, smallest validation, non-goals, and a stop or revise rule. Use when a PM must choose what to pursue next without turning an unsourced score, single signal, or market guess into a roadmap commitment.

asdc163 Bundle

Use when an AI capability may be packaged as a reusable agent skill. Produce a source-bounded package contract covering identity, discovery triggers, progressive disclosure, resources and scripts, permissions, surface compatibility, provenance, versioning, verification, disablement, rollback, and a truthful release decision.

asdc163 Bundle

Use when a PM needs to decide when an AI capability should be surfaced, selected, declined, or handed to a manual route based on user intent, context, host capability, and permission.

asdc163 Bundle

Turn scattered notes, workflow artifacts, tests, adoption observations, metrics, and stakeholder feedback into a source-bounded AI workflow evidence packet. Separate what is measured, observed, reported, estimated, planned, inferred, or unknown; preserve contribution and conflict; and choose Capture next, Proceed bounded, Hold, or Do not claim without inventing adoption, value, causality, safety, or production readiness.

asdc163 Bundle

Use when an AI agent has a large or changing tool catalog and a product manager must define what is searchable, which candidates are eligible, when tools are deferred or loaded, how hosted and client-owned discovery differ, when the agent should abstain, and how selection stays separate from authorization, execution, outcome, and adoption.

envoydev Skill

.NET Aspire conventions for local cloud-native orchestration - the AppHost that declares the topology, the shared ServiceDefaults extension every service calls once, name-based service discovery, AppHost-injected connection strings, and the developer dashboard. This is the local run, NOT production deployment or container publishing. Floors at .NET 8 / C# 12. Load when scaffolding or editing an AppHost or ServiceDefaults, declaring resources and references, wiring discovery, or when the user says Aspire, AppHost, AddProject, WithReference, service discovery, or Aspire dashboard. Companions: dotnet-web-backend, dotnet-testing. Do NOT load for non-Aspire projects, production deployment, or publishing images.

kaiohenricunha Bundle

Grade implementation plans on a 0-10 scale without mutating the workspace. Accepts plan files, quoted inline plans, or best-effort latest-plan discovery for Claude, Codex, Copilot, and Gemini.

jperezduerto Skill

Operate ffuf (Fuzz Faster U Fool) — an extremely fast web fuzzer written in Go. Use when the user needs to brute-force web directories, discover virtual hosts, fuzz GET/POST parameters, headers, or subdomains, or when the user asks about ffuf, directory brute forcing, web fuzzing, or content discovery. Covers installation, all filtering and matching flags, multiple wordlist positions (FUZZ, custom keywords), VHOST discovery, parameter fuzzing, recursion, output formats, rate limiting, authentication (cookies, headers), custom wordlists, and integration with SecLists.

jperezduerto Skill

Build, extend, and operate nmap — the industry-standard network mapper and port scanner. Use when performing host discovery, port scanning, service enumeration, OS detection, or running NSE scripts during penetration testing engagements or network audits. Covers host discovery techniques, scan types (SYN, TCP, UDP, version, script), timing templates, NSE scripting categories, output formats, firewall evasion, performance tuning, and integration with Metasploit db_nmap and downstream tools.

jperezduerto Skill

Build, extend, and operate Amass — the OWASP attack surface mapping tool for subdomain enumeration, ASN/org discovery, and DNS intelligence. Use when performing external reconnaissance, passive or active subdomain enumeration, certificate transparency analysis, ASN mapping, or tracking changes in an attack surface. Use when the user asks about subdomain discovery, passive DNS, API key integration, graph visualization, Amass config files, or comparing Amass with subfinder. Covers installation, all subcommands, config, data sources, brute force, resolvers, and visualization.

jperezduerto Skill

Operate Arjun — an HTTP parameter discovery tool that finds hidden GET, POST, JSON, and XML parameters in web endpoints through intelligent wordlist-based bruteforcing. Use when mapping an application's attack surface, discovering undocumented API parameters, or preparing targets for fuzzing with ffuf, SQLMap, or XSStrike. Covers installation, GET/POST/JSON/XML modes, Burp import, custom wordlists, threading, output formats, and integration into a full web application testing workflow.

jperezduerto Skill

Build, extend, and operate Masscan — the fastest Internet port scanner, capable of scanning the entire IPv4 address space in under 6 minutes at 10 million packets/second. Use when performing large-scale network reconnaissance, port enumeration, banner grabbing, or building scanning pipelines. Use when the user asks about high-speed scanning, comparing Masscan vs nmap, rate tuning, output formats, or integrating Masscan with nmap and Nuclei for vulnerability discovery pipelines.

jperezduerto Skill

Operate RustScan — an extremely fast modern port scanner that scans all 65,535 ports in seconds and automatically passes results to Nmap for deep service/script scanning. Use when the user needs fast port discovery, when the user asks about RustScan, when conducting external or internal network reconnaissance, or when standard Nmap scans are too slow. Covers installation (cargo, Docker, package managers, releases), batch size tuning, timeout configuration, Nmap integration, custom scripts, accessible mode, performance tuning, and how RustScan compares to and chains with Nmap and masscan.