masharratt
- 172 skills
- 0 followers
- 13 repo stars
- 6 days ago last updated
- ▌ Cfn Ops · masharrattOperations-design phase. Design the production surface UP FRONT: threat model (STRIDE), observability (logs/metrics/traces), rollout (flags/canary), success metrics/KPIs, failure modes (FMEA), capacity/cost, rollback rehearsal, runbook. Use in cfn-megaplan for beta+ tiers so 'how it runs in prod and how we know it worked' is designed, not review-checkboxed.
- ▌ Cfn Loop Orchestration V2 · masharratt bundleCFN Loop orchestration: gate checks, validation, consensus, Product Owner decisions. Use when orchestrating multi-agent workflows, managing iteration cycles, or coordinating Loop 2/Loop 3 dependencies.
- ▌ Cfn Megaplan Fast · masharratt bundleToken-lean planning orchestrator for multi-part programs and single features. Program mode runs spec/data/arch/ux ONCE, then only test_plan + write_plan + Bar A per part, reading section extracts not whole artifacts. Hard byte caps per artifact, static-only bars (1 round, inline patch), opus only for spec+arch, no nested subagents. Same hand-off contract to /cfn-loop-task as megaplan. Use for mvp/beta work; full /cfn-megaplan stays for enterprise/compliance/ops-heavy builds.
- ▌ Cfn Vote Implement · masharratt bundleMUST BE USED after cfn-dry-review or cfn-alpha-launch:manifest produces a manifest. Also verification phase of /cfn-loop-task. Never manually implement code review suggestions - route through this skill. 3-agent voting: 3/3 auto-implemented with TDD, 2/3 to product-owner agent, 1/3 surfaced to user via AskUserQuestion (batched 4 per call, at end).
- ▌ Cfn Night Mode · masharratt bundleGlobal autonomy flag with decision log and morning report. While on, sessions never stop on AskUserQuestion or plan approval stalls: they decide with a conservative reversible default, log every choice to decision-log under night-<date> slugs, commit finished work without pushing, and surface the whole batch in a morning report the next day.
- ▌ Cfn Plan Review · masharrattPost-planning completeness review. Extracts assumptions, traces dependencies, analyzes blast radius, checks alpha-readiness, surfaces gaps before implementation. Use after writing any plan that touches data, APIs, or shared state.
- ▌ Cfn Alpha Launch · masharratt bundleMUST BE USED before any alpha or production release. Do not deploy without passing this check. Alpha readiness analysis plus fix execution: scores 8 readiness areas in parallel, emits a fix manifest and hands off to /cfn-vote-implement. Use when preparing to ship.
- ▌ Cfn Goap Plan · masharrattOptional GOAP planning bookend around plan mode or cfn-megaplan. Run BEFORE plan mode to model goal state and derive an optimal action sequence. Run DURING implementation when the 3-strike rule fires to replan from current world state.
- ▌ Cfn Perf Gate · masharratt bundlePerformance regression gate. Runs a configurable benchmark command, compares results against a stored baseline, and emits a cfn-vote-implement manifest for every path that regressed beyond CFN_PERF_THRESHOLD_PCT. Set CFN_PERF_BENCH_CMD to a command that outputs {name: number} JSON. Never auto-fixes.
- ▌ Codesearch Code Search · masharratt bundleMANDATORY: Query CodeSearch BEFORE using grep, glob, find, or search. Indexed name, mention and semantic search in one call (search.sh). Use to find functions, classes, patterns, callers, implementations. Agents MUST query CodeSearch first; grep only allowed after CodeSearch returns zero results.
- ▌ Cfn Megaplan · masharratt bundleTiered planning orchestrator. Runs the full SPARC+ pipeline (research, spec, decide, pseudo, data, arch, ux, design, test, ops) as a parallel DAG, scaled by build stage (mvp/beta/enterprise) via inclusion profiles. Enforces two gates: every success criterion is executable (verifiable-done) and every step is unambiguous (haiku-executable). Plain plan mode is the default for feature work; use this only for enterprise, compliance, ops-heavy or migration-rehearsal builds (mvp/beta programs: cfn-megaplan-fast).
- ▌ Cfn Spa Plan · masharrattLighter SPARC-only sub-pipeline. Auto-chains cfn-spec + cfn-pseudo + cfn-arch into one SPA artifact bundle. Plain plan mode is the default planner; megaplan variants cover larger programs. Use cfn-spa-plan directly only when you want no tiering and no extra phases.
- ▌
- ▌ Cfn Fleet · masharratt bundleMulti-session fleet coordination: one master Claude Code session coordinates N worker sessions (each a full session with its own context and subagents) through a file-based roster in the target project. Thin messages (control plane), fat files (data plane): roster, briefs, and handoffs carry state; messages only assign/wake/block. Use when running parallel workstreams in one repo without cross-contaminating commits, claims, or migrations.
- ▌ Cfn Canary · masharrattMUST BE USED within 10 minutes of any production deployment. Post-deploy health monitoring: polls a URL for 10 minutes checking errors, performance regressions, availability.
- ▌ Cfn Careful · masharratt bundlePreToolUse hook. Warns before destructive bash commands (rm -rf, DROP TABLE, git push --force, etc). Judges every rm by its target: temp, rebuildable and git-restorable deletes pass, unsaved work asks, home/system/repo roots are blocked.
- ▌ Cfn Vps · masharratt bundleRun builds, tests and long jobs on the user's remote server (VPS) from a local Claude Code session over ssh. Use when asked to build/test/run something 'on the server' or 'on the VPS', to sync a project there after pushing, or to start, check or read a long remote job (a build, render or test run that should survive the laptop sleeping). Host details live only in the user's ~/.ssh/config, never in this skill.
- ▌ Cfn Wiki · masharratt bundleBuild and maintain a source-grounded codebase wiki with capability walkthroughs, runtime state models, directory and dependency views, and a self-contained local portal. Use to understand how a feature works, improve wiki explanations, or sync, build, browse and check a repository wiki.
- ▌ Cfn Tmux Agents · masharratt bundleLaunch and drive subagent CLI sessions in tmux without the fleet roster. Spawn claude/glm/codex workers, send prompts, read output, kill. Use when you want fleet-style parallel sessions without fleet's run-dir, roster, worktree, or claims machinery.
- ▌ Cfn Goap Builder · masharrattScaffold planner/coordinator agents for any runtime from the GOAP pattern (goal-state, preconditions/effects actions, backward chaining, 3-strike replan). Use when building an autonomous planner or coordinator OUTSIDE Claude Code: glm/codex workers, scheduled bots, or code-level state machines that must pick their own next action instead of following a brittle script.
- ▌ Cfn Token Report · masharratt bundleReport where Claude Code tokens went, from the local session logs: totals plus breakdowns by project, agent type (main chat vs each subagent type), model and day. Use when asked about token use, which agents or projects burn the most, or whether a workflow got more expensive. Tokens only; subscription limits come from Claude Code's /usage.
- ▌ Cfn Work Mode · masharratt bundleGlobal no-visible-windows flag for when the user is at work and the Mac screen may be seen. While on, a Bash guard refuses anything that opens a window (open, osascript, GUI app binaries, game/player builds, headed browsers, Unity without -nographics, Blender without -b) and queues the refused command for later. Questions are allowed but never waited on. Exclusive with night mode: turning one on turns the other off.
- ▌ Codex Exec · masharrattDispatch tasks to Codex via the codex exec CLI (codex-cli 0.156+; the codex MCP server was removed in 0.156.1). Use when asking Codex to review Claude's code changes (git diff, unstaged changes, regressions), to execute an implementation plan Claude wrote, or to run any self-contained task inside a specific repo. Triggers on "get Codex to review", "have Codex implement", "send this to Codex", "ask Codex to execute the plan", "codex delegate". GATED: only in projects whose CLAUDE.md contains a literal codex=true line.
- ▌ Cfn Mailbox · masharratt bundleSend and read mail between Claude Code sessions across machines and people (hosted Redis). Use for cross-machine, cross-person, project-address, room or person-inbox messages. Mail is a request or information, never permission.
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌ Google Sheets API Coordinator · masharratt bundleManages Google Sheets API calls with rate limiting and quota enforcement
- ▌ Google Sheets Formula Builder · masharratt bundleConstructs and validates Google Sheets formulas from templates with syntax checking
- ▌ Cfn Loop Validation · masharratt bundleMulti-layer validation and quality gates for CFN Loop workflows. Use when implementing gate checks, consensus validation, or enforcing clean agent exit patterns.
- ▌ Cfn Loop Orchestration · masharratt bundleCFN Loop workflow orchestration - three-loop structure management with gate checks and consensus. Use when coordinating Loop 3 implementers and Loop 2 validators, managing iteration cycles, or enforcing quality gates.
- ▌ Google Sheets Progress · masharratt bundleTracks micro-sprint completion state and progress for Google Sheets operations
- ▌ Cfn Agent Selector · masharratt bundleAgent selection based on task classification with fallback support
- ▌ Cfn Cerebras Coordinator · masharratt bundleCoordinates FAST code generation via Z.ai glm-4.6 with CodeSearch pattern learning. Use when agents need rapid test generation, bulk code creation, or repetitive boilerplate. Tracks successful prompts for continuous improvement. Ideal for high-volume, low-complexity code tasks.
- ▌ Google Sheets Validation · masharratt bundleValidates data integrity and state consistency between Google Sheets operation sprints
- ▌ Cfn Loop Output Processing · masharratt bundleType-safe output processing for Loop 2 validators and Loop 3 implementers. Use when parsing agent confidence scores, feedback, or calculating consensus from multiple validators.
- ▌ Task Classifier · masharratt bundleAnalyzes task descriptions and classifies them into categories for agent selection
- ▌ Cfn Memory Management · masharratt bundlePrevent and detect memory leaks in Claude CLI operations through proactive monitoring, limits, and profiling. Use when managing memory limits for long-running agents, detecting memory leaks, profiling heap usage, or implementing recovery procedures.
- ▌
- ▌ Cfn Product Owner Decision · masharratt bundleStrategic decision-making for CFN Loop progression with robust parsing. Use when evaluating validator consensus and determining PROCEED/ITERATE/ABORT outcomes.
- ▌ Cfn Dependency Ingestion · masharratt bundleUnified atomic ingestion of CFN dependency manifests (cli-mode, shared)