Every prompt pack in the store.

AI Agent Acceptance Testing Harness Playbook

Put any AI agent through real user-acceptance testing before it touches production — baseline its job, build a scenario matrix with edge cases, sequence a risk-ordered rollout with rollback triggers, and stand up an ongoing sign-off cadence.

Agent Commit Security Harness

Stop agent-authored commits from leaking secrets or introducing unsafe patterns. Four specialized prompts scan every diff for credentials, injection flaws, auth mistakes, and risky dependencies — then post a prioritized security report to your PR. Requires GitHub and Slack connected to your agent.

Coding Agent Eval Harness Builder Playbook

Stand up a repeatable golden-task evaluation harness for your coding agent — define representative tasks, scoring criteria, CI wiring, and regression tracking in one structured playbook.

Agent Fleet Operations Playbook

Turn a sprawl of 5–50 AI agents into a governed fleet: a criticality-classified inventory with intake rules, per-agent quality gates sized by blast radius, cost budgets with alert ladders, and a weekly review where every agent ends in a keep, fix, throttle, expand, or retire decision.

Agent Fleet Supervisor Daily-Ops Kit

Run a real supervisor shift over a fleet of AI agents: paste last night's run export and get a triage board that catches silent green failures, a capacity-bounded exception queue with containment moves, drift and escalation calls against explicit thresholds, and a handover note whose carry-forward block starts tomorrow's shift.

AI Guardrail Bypass Red-Team Kit

Systematically probe your AI guardrail implementation before it hits production or audit. Five composing prompts generate a full attack taxonomy, targeted bypass scenarios, multi-turn escalation scripts, an evidence capture template, and a prioritized remediation plan — so you can demonstrate robustness to regulators, not just hope for it.

Coding Agent Guardrails System Prompt

Install a production-grade guardrail policy into any coding agent — forbidden paths, change-size caps, mandatory self-review, and human escalation rules — so it stays safe on your real repo.

Agent Identity & Access Designer Playbook

Design the identity model for your AI agents before you provision them: assign every agent action a principal archetype (dedicated service identity, delegated human identity, dual principal, per-tenant), write six-axis boundary contracts with blast radius statements, plan credential issuance and rotation with joiner-mover-leaver coupling, and finish with an attribution field contract and a rehearsed break-glass revocation runbook.

Agent Memory Architecture & Audit Playbook

Diagnose and audit your LLM agent's memory the way a specialist would — isolate why it forgets, recalls the wrong thing, or bloats, score the setup across four failure modes, and get a prioritized fix list. Includes the design prompts to architect working, episodic, semantic, and procedural memory from scratch. No connected tools — runs in any chat model.

Agent Code Output Verification Rubric

Score agent-produced code changes against correctness, safety, test coverage, and convention adherence — and get a weighted merge verdict before anything reaches your main branch.

LLM Agent Output PII and Secret Leak Scanner

Systematically audit every surface where your LLM agent can leak PII or secrets — reasoning traces, tool call payloads, logs, and final responses — and produce a compliance-ready redaction policy your AppSec team can hand to a GDPR or HIPAA auditor.

Agent Prompt Regression Rubric

Detect and score regressions when you change a production agent prompt — baseline expected behavior, compare new vs old on representative tasks, classify every regression, and get a ship or hold verdict.

Agent ROI Measurement Rubric

Build a defensible ROI case for any AI agent initiative — honest baseline, counterfactual cost model, quality-adjusted throughput score, and an exec scorecard with a continue/expand/kill call.

Agent Operations Runbook Generator

Turn any agent or automation description into a complete operational runbook — failure mode catalog, step-by-step diagnostic guides, recovery and rollback procedures, and a clear escalation matrix — so on-call is never unprepared when automations break.

Autonomous Agent Sandbox Setup Playbook

Design a safe, isolated execution environment for autonomous coding agents — container boundaries, permission models, CI feedback loops, and kill-switch procedures in one playbook.

Agent Task Decomposition System Prompt

Stop your coding agent from making sweeping, unreviewable changes to large repos. This system-prompt harness forces decomposition-first execution: scope mapping, dependency ordering, checkpoint verification, and human-gated stop conditions — before a single file is edited.

AI Agent Least-Privilege Auditor

Audit every tool your AI agent holds against least-privilege principles: extract a complete permission inventory, score each excess grant by blast radius and compliance exposure, map gaps to your RBAC policy, and generate an auditor-ready evidence pack — in one structured workflow.

Agent Vendor & Tool Evaluation Rubric

Score any AI agent product or MCP server before you buy: build a weighted criteria model, separate verified facts from vendor claims, produce a scored Adopt/Pilot/Hold/Reject verdict, and rank the open risks into a procurement action plan. Works in any chat model — no connected tools needed.

AGENTS.md Generator for Polyglot Monorepos

Turn a raw repo tree and stack summary into a hierarchical AGENTS.md / CLAUDE.md set that gives every coding agent accurate, per-directory operating rules — eliminating misbehaving agents and one-off instruction rewrites.

AI Agent Audit Trail Designer Pack

Design tamper-evident, regulator-ready audit trails for autonomous AI agents. Five structured prompts produce a log schema, regulatory traceability matrix, retention policy, tamper-evidence architecture, and incident replay template — so your next regulator call is answered with proof, not apology.

HR·Intermediate

AI-Era Workforce Planning Framework

Plan a team whose work is partly done by machines: score every work item on realizable exposure, re-forecast capacity after the verification tax and demand rebound eat the freed hours, find which skills gain and lose leverage, rewrite your open hiring profiles, and sequence the reskilling waves that close the gap.

AI Feature Spec System Playbook

Write a shippable spec for a probabilistic AI feature — acceptance thresholds, fallback UX, eval criteria, and a failure budget — plus a risk-tiered rollout and an error-budget governance loop. The parts a normal PRD leaves blank because the feature can be confidently wrong. Built for PMs and AI engineers who ship AI features together.

AI Incident Commander Runbook System

When your autonomous AI agent DOES something wrong in production — sends, charges, deletes, deploys — run the live command loop: classify severity from what was actually executed, throttle the agent's autonomy, walk the failure-class diagnosis tree, notify the people it touched, and close out into a cross-incident prevention ledger. Covers agent-action incidents; output-quality LLM incidents belong to the LLM Incident Response Runbook System.

AI-Output Auditor Rubric Kit

Run a defensible human review program over AI-generated deliverables: calibrate a weighted six-dimension rubric to your output class, decide how much to sample, score each deliverable to a ship-or-revise verdict with evidence, and roll the scorecards up into a root-cause fix plan.

AI PR CI-Gate Harness Agent Pack

Stop AI-generated PRs from slipping policy violations or scope creep into your codebase — a four-prompt harness that classifies risk, checks compliance, detects scope drift, and posts a structured approve/block decision to GitHub automatically.

AI Red-Team Operations Kit

Run adversarial testing of any LLM feature as a governed program, not a one-off. Five composing prompts calibrate a severity-and-confidence scoring rubric, grade raw red-team logs into verified evidence, produce a scored findings report, prioritize remediation by risk versus effort, and lock confirmed exploits into a re-runnable regression suite.

AI Usage Policy & Governance Playbook

Write an AI policy employees actually follow: map the real AI risk surface including shadow usage, draft enforceable plain-language rules around a data-tool matrix, align with EU AI Act categories and SOC 2 / ISO controls, and roll out with amnesty framing, a fast exception process, and proportional enforcement.

API Contract Test Harness Pack

Generate contract tests that catch breaking API changes between producer and consumer services — extract the contract, map consumer expectations, produce failing tests, and get safe versioning guidance.

API Design Review Evaluation Rubric

Put every API design through the same weighted rubric before consumers depend on it: scored dimensions for consistency, usability, versioning, scalability, and operational impact — each backed by cited evidence from your spec — ending in a prioritized fix list ordered by consumer pain, not reviewer taste.

Architecture Decision Record System

Stand up a working ADR practice for your team: diagnose which decisions are undocumented, author complete decision records with forced tradeoff scoring, reversibility ratings, and captured dissent, then sequence the backlog and run a recurring decision-review cadence. Works in any chat model — no connected tools.

Automation Portfolio Manager OS

Govern your whole automation estate — zaps, n8n and Make workflows, cron jobs, Airflow DAGs, RPA bots and agents — like an investment book: a classified register that surfaces duplicates and orphans, every asset scored on value versus true maintenance cost, dated retire/consolidate/invest calls that fit your real capacity, and a quarterly review with named owners. Works in any chat model; paste your platform exports, no tools required.

NeedsJiraSSlackConfluence

Productivity·Advanced

Backlog Cleanup Agent Pack

Turn a sprawling Jira backlog into a prioritized, groomed work queue — duplicate tickets merged, priorities realigned to business goals, acceptance criteria filled in, and a stakeholder health report published automatically.

Board Pack Production System

Produce a board-grade pack in a day instead of a week: a three-paragraph headline narrative, plan-vs-actual bridges with driver attribution, the hardest director questions pre-answered, and a disciplined pack outline with per-slide message titles.

Breaking Change Detection Harness Agent Pack

Diff your library's public surface before every release, classify what breaks consumers and why, recommend a semver bump, write migration notes, and attach everything to the release PR automatically. Requires GitHub connected to your agent.

Build System Unification Playbook

Audit every build tool in your monorepo, pick the right unified system, sequence the package migrations, and design remote caching — ship a CI transformation plan that actually sticks.

Build-vs-Buy Decision Engine

Settle any build-vs-buy call with a total-cost-of-ownership model, a weighted capability-fit scorecard, a lock-in and exit read, and a decision memo your leadership will actually approve.

HR·Advanced

NeedsGoogle SheetsNotion

Candidate Screening Agent Pack

Screen every applicant against the same evidence-based scorecard: criteria-anchored ratings from a connected Google Sheets list, an objectively ranked shortlist with cited reasons, targeted interview questions per gap, and a hiring-team brief in Notion — as consistent for candidate one hundred as for candidate one.

Champion Development Playbook

Stop calling every friendly contact a champion. Score each named person on observable behaviour, build the internal business case in the account's own vocabulary, rehearse the objections finance, security and their manager will actually ask, sequence commitment asks that test willingness without pressure, and score how exposed the deal is if that one person leaves or goes quiet.

CI Failure Diagnosis Harness Agent Pack

Cut the time engineers spend decoding red CI from hours to minutes — a four-prompt harness that reads pipeline logs, separates real failures from infra flake, recommends the concrete fix, and posts the diagnosis to GitHub and Slack automatically.

Code Review Policy System Prompt

Give every reviewer — human or agent — the same consistent review standard: a four-prompt system prompt set covering review dimensions, severity tiers, repo-specific rules, and actionable comment format, so no PR slips through with vague feedback.

Codebase Onboarding Harness Agent Pack

Drop an AI agent into any unfamiliar polyglot repo and get a production-ready onboarding map — architecture, entry points, build commands, and a safe first task — in minutes, not days. Requires GitHub connected.

Codemod Authoring Harness

Author and safely apply codemods that transform code consistently across a polyglot repo — a four-stage harness that specifies, generates, dry-runs, and verifies every mechanical change before it lands.

Commercial Lease Risk Review Engine

Read any commercial lease like a real-estate attorney would: extract every economic term, run the NNN/CAM math, score each clause against market norms, and walk away with a prioritized redline list and a landlord-ready negotiation message.

Documentation Maintenance Agent Pack

Keep Confluence documentation accurate as projects evolve — audit stale pages, detect which docs are affected by recent Jira work, draft precise updates, and publish with Jira and Slack close-out in one connected workflow.

Context Engineer Role Playbook

Run context engineering as a standing discipline for your agent fleet — build the team's source-of-truth register, publish the retrieval policy and token budget standards every workflow follows, and catch context rot with a quarterly audit.

Context Engineering Harness

Run context as an engineering discipline, not guesswork — inventory and rate every source your agents consume, write a per-job loading policy, fit it to a hard token limit with a cut log, and audit the setup for stale or contradictory context.

Contract Risk Triage Rubric

Triage inbound contracts in an hour instead of a week: extract every operative clause with definition chasing and silence detection, score deviations against your own positions with consequence scenarios sized to the deal, build capped negotiation ladders with trades, and get a sign/negotiate/escalate verdict. Preparation for counsel — not legal advice.

Marketing·Intermediate

CRO Audit Engine

Stop testing random ideas — diagnose exactly where your funnel loses buyers with a friction taxonomy, audit trust signals against what your audience needs, and walk away with a test backlog ranked by expected value plus implementable test specs.

Cross-Language Refactor Harness

Coordinate a dependency-aware refactor across multiple languages without breaking call sites — a disciplined four-stage harness that maps, sweeps, guards, and verifies every change your agentic editor makes.

NeedsGitHubConfluence

Cross-Service Dependency Audit Agent Pack

Map every service-to-service dependency, flag cycles and tight coupling, and publish a risk-ranked decoupling plan to Confluence — all driven by your connected GitHub and Confluence agent.

NeedsSentryGitHubSSlack

Cross-Stack Error Correlation Agent Pack

Turn scattered multi-service errors into a single root-cause story — aggregate Sentry errors across services, stitch them along request traces, propose the originating failure, and file a prioritised issue in one agent session.

Customer Support·Advanced

NeedsZendeskSSlackJira

Customer Escalation Resolution Agent Pack

Resolve critical customer escalations faster — assess severity, assign investigation tasks in Jira, draft customer-ready updates, and write post-resolution documentation, all from one connected agent workflow.

NeedsIntercomNotionSSlack

Customer Feedback Synthesis Agent Pack

Turn scattered customer feedback from Intercom into prioritised product insights — cluster themes, score business impact, and publish a stakeholder-ready summary to Notion and Slack automatically.

Customer Support·Advanced

NeedsHubSpotSSlackGoogle Sheets

Customer Health Monitor Agent Pack

Spot at-risk accounts before they churn — read engagement signals from HubSpot and Google Sheets, classify risk, build retention plans, and push Slack alerts to your CS team automatically.

Cybersecurity Tabletop Exercise (TTX) Builder

Design and run a full incident-response tabletop exercise without a $5K-$50K consultant: a threat-tailored scenario, a timed inject sequence with real decision points, role and RACI assignments, a minute-by-minute facilitator run-sheet, and a NIST-scored after-action report leadership can sign off on.

Data Quality Incident Harness Playbook

When wrong numbers have already reached dashboards, board decks, invoices and trained models, run the data-incident loop end to end: classify the failure and bound the corruption window, trace which consumers the bad values reached and where they are now frozen, decide per audience what must be formally restated versus quietly corrected, sequence a gated backfill with real verification, draft the correction notice for people who already acted, and close the detection gap that let it run silent.

Database Migration Safety Harness

Vet any schema migration for lock, downtime, and data-loss risk before it ships — then get a safe rollout plan and a tracked GitHub issue, all from one agent workflow.

Department Automation Roadmap Playbook

Answer "what should we automate with AI?" with evidence instead of anecdotes: a complete scored inventory of your department's workflows, an anchored automation-fit ranking with kill conditions, a sequenced two-quarter roadmap, and approval-ready initiative briefs with honest ROI and pre-committed kill criteria.

Dependency Upgrade Harness Agent Pack

The judgment layer Renovate and Dependabot are missing: inventory outdated dependencies across every service, assess each upgrade's real blast radius from changelogs and usage, batch the safe ones, sequence the dangerous ones, and open verified upgrade PRs with rollback notes. Requires GitHub connected to your agent.

Research·Intermediate

Discovery Interview Synthesis Engine

Turn a backlog of interview transcripts into a decision-ready opportunity tree — atomic claims with verbatim evidence, cross-interview themes scored by frequency and severity, and a stakeholder readout that forces a call instead of a nod.

Docs-Code Sync Harness Agent Pack

Keep developer docs accurate as code evolves — map docs to source, detect stale sections after every merge, draft precise edits, and open a reviewed PR automatically. Requires GitHub connected to your agent.

Eng Org Design Framework

Redesign your engineering org with practitioner judgment built in: a cognitive-load audit, Team Topologies scoring, a 12-mode reorg pre-mortem, and a ready-to-ship decision memo plus comms sequence.

HR·Advanced

Engineering Hiring Loop Designer

Diagnose your broken hiring process against a 12-entry failure-mode catalog, then design and roll out a full structured loop — role scorecard, stage-to-signal architecture, interviewer certification ladder, and a calibration cadence that catches score drift before it costs you a hire.

Engineering Metrics Interpretation Engine

Turn DORA, SPACE, and cycle-time data into an honest engineering narrative and intervention plan — with confound checks, metric-gaming detection, and a leadership memo that survives scrutiny. Works in any chat model, no connected tools.

NeedsPagerDutySSlackJiraDatadog

Engineering On-Call Coordination Agent Pack

Coordinate the full on-call incident lifecycle — alert triage, responder notification, stakeholder status updates, and Jira follow-up tracking — across PagerDuty, Slack, Datadog, and Jira without leaving your agent.

Engineering Strategy OS

Turn a reactive roadmap into a written engineering strategy: an honest current-state diagnosis, a Rumelt-style guiding policy with coherent actions and a scored tech-bet portfolio, a risk-and-dependency-sequenced rollout, and a governance cadence with owners, guardrails, and a comms plan. Works in any chat model — paste your context, no tools required.

Enterprise Account Research Agent Pack

Turn a company name into a sourced account dossier your team can defend in front of the buyer: stated priorities and trigger events pulled from filings and earnings calls, a tech-stack and incumbent read, a buying-committee map with unfilled roles named, and a point-of-view brief where every claim carries its source, tier, and confidence. Built for a browsing-enabled agent; degrades to pasted source material when no tool is connected.

Environment Parity Harness Pack

Detect dev, CI, and prod environment drift, pinpoint the differences causing 'works on my machine' bugs, and get a concrete parity plan with guardrails to prevent it from recurring.

Eval Engineer Career Toolkit

Run eval engineering for your agentic workflows as a documented, governed program — a coverage-designed golden set, audited graders, a regression gate policy with a real flake budget, and a drift runbook with re-baselining rules. (Operates the practice week over week; for one-time eval architecture of an LLM feature, see the LLM Eval System Design Playbook.)

Experiment Design Review Rubric

Review a proposed A/B test before it gets traffic: pin the design down, run honest power and interference math with the assumptions attached, score it on a weighted validity rubric with hard gates, and hand back the required fixes plus a filled pre-registration record. Works in any chat model — no connected tools needed.

NeedsPostHogNotionSSlack

Feature Adoption Analysis Agent Pack

Turn PostHog usage data into a prioritised adoption report — analyse behaviour patterns, compare segments, pinpoint drop-off, and publish recommendations to Notion and Slack. Requires PostHog, Notion, and Slack connected to your agent.

NeedsGitHubJiraSSlack

Feature Flag Governance Agent Pack

Audit your entire feature flag inventory, assess operational risk, generate prioritized cleanup recommendations, and publish a governance report — all driven by your connected GitHub, Jira, and Slack tools.

Connected Design Handoff Agent Pack

Convert approved Figma screens into implementation-ready engineering artifacts — extract specs, write detailed requirements, create structured Jira tickets, and publish a handoff brief to Slack. Requires Figma, Jira, and Slack connected to your agent.

Fine-Tune vs RAG vs Prompt Decision Engine

Settle the fine-tune vs RAG vs prompt-engineering question with a weighted cost/latency/maintenance scorecard, a reversibility and failure-mode read, and a decision memo your team will actually approve.

NeedsGitHubJiraSSlack

Flaky Test Detection Harness Agent Pack

Detect, root-cause, and quarantine flaky tests across your CI pipeline — then automatically create tracked fix issues so unreliable tests stop blocking good changes. Requires GitHub, Jira, and Slack connected to your agent.

Framework Migration Harness Playbook

Inventory every service, map every breaking change, sequence your rollout, and define safe rollback paths — ship a complete framework-migration plan your team can execute with confidence.

Freelance Contract & Scope-Creep Shield

Stop signing away your IP and doing free out-of-scope work: scan any client contract for the IP, portfolio, payment, and indemnity traps boilerplate hides, get send-ready redlines, lock a proposal into an in-scope/out-of-scope baseline, and turn every 'can you just...' request into a priced change order. Preparation, not legal advice.

Strategy·Advanced

Fundraising Narrative Playbook

Turn raw company facts into a fundable seed-to-Series-B raise: a why-now/why-you/why-this-wins narrative with graded evidence, honest metric framing, a 20-objection bank with rehearsed answers, and a sequenced investor process with a debrief loop that compounds every meeting.

Marketing·Intermediate

Funnel Diagnostics Engine

Diagnose before you optimize: normalize your funnel data, localize the leaks that cost the most recoverable value, rank cause hypotheses across message, audience, friction, and trust — then design the decisive experiment for each.

~40 min to use

Golden-Path Service Scaffolder

Scaffold any new service to your repo's golden-path standards — directory layout, config, CI stub, and registration checklist — without tribal knowledge or manual drift.

GTM Motion Selection Framework

Decide between product-led, sales-led, hybrid, and partner-led selling with the constraints that actually settle it: deal value against cost to serve, whether your product delivers value before a human touch, how many buyers and how much procurement stand in the way, and payback measured against your runway — then stress test the winner against ten named failure modes and leave with a staged test plan that proves the motion before you hire against it.

Hallucination Failure-Mode Audit Rubric

Audit any LLM feature against a 25-class hallucination taxonomy — run per-class detection probes, score every finding by severity and likelihood, and ship a prioritized per-feature risk register and remediation plan.

Human-in-the-Loop Design Playbook for AI Agents

Decide exactly where humans belong in your AI agent workflow — a risk-tiered map of approval gates, sampling rates, escalation triggers, and reviewer rubrics that stops both rubber-stamping and review bottlenecks. Four prompts take you from current-state diagnosis to a governed, rollout-ready oversight design.

Hypothesis-Driven Analysis Playbook

The consulting working method as a system: sharpen a vague problem into a governing question, decompose it into a MECE issue tree with kill conditions, prioritize falsifiable hypotheses against real capacity, run minimal decisive analyses with pre-committed decision rules, and synthesize the answer as a pyramid.

Infrastructure Drift Detection Harness

Detect IaC drift before it causes outages — parse declared state from your repo, compare against observed reality, score each deviation by blast radius, and get a safe reconciliation plan with tracking issues ready to open. GitHub connected unlocks direct repo reads and automatic issue creation; every step also works by pasting file or state content directly.

ICP Discovery Engine

Derive your ideal customer profile from your own closed-won and churned accounts instead of inventing a persona: segment the book you already sold, separate attributes that actually predict retention from coincidences, name the disqualifiers your reps can check on a first call, and leave with a weighted fit score that has been back-tested against known outcomes and a falsification test for next quarter.

NeedsPagerDutyJiraSSlackConfluence

Incident Postmortem Agent Pack

Turn raw incident signals from PagerDuty, Jira, and Slack into a complete postmortem — accurate timeline, root cause analysis, Jira remediation tasks, and a Confluence-published executive summary — without the manual multi-system grind.

Incident Review Quality Rubric

Grade any written postmortem before you accept it: a weighted five-dimension scorecard with 1-5 anchors, a shallow-analysis and blame detector, an action-item verifiability audit, a reconstructed detection clock, and a section-by-section rewrite brief you can send straight back to the author.

Insider Threat Program Builder

Stand up a documented insider-threat program your cyber insurer or auditor will accept — a CISA IRMPE maturity score, a controls plan mapped to your tools, a detection catalog, a response runbook, and a policy plus insurance-evidence pack. No specialist consultant required.

Internal Agent Adoption Playbook

Turn a successful AI-agent pilot into org-wide daily usage instead of a stalled experiment: a scored adoption diagnostic with a named resistance map, a champion-network and workflow-embedding system, a risk-sequenced rollout, and a governance cadence with usage instrumentation and pre-committed expand/kill gates.

Investment Thesis Construction OS

Turn a folder of research notes into a thesis you can actually be wrong about: one falsifiable core claim, the mechanism chain behind it, base-rate reality checks on every link, the disconfirming evidence you would need to see, numeric kill criteria with action tiers, and a monitoring schedule that tells you when the thesis has broken. A structuring system for your own analysis — not investment advice.

~90 min to use

IP Ownership Gap Audit: Do You Actually Own Your Product?

Find out whether you truly own your codebase and IP before diligence does. Inventory every contributor, test each one for a valid IP assignment, get the ownership-gap report a VC or acquirer would write, and a fix-it kit with assignment-request messages and clause language.

Legacy Code Comprehension Harness

Map, explain, and safely change legacy multi-tech systems before the last person who understood them leaves. Connected agents read the code directly and publish findings to Confluence so nothing is lost. Requires GitHub connected; Confluence optional.

NeedsLinearSSlackNotion

Release Readiness Agent Pack

Verify milestone completion in Linear, surface unresolved blockers and dependencies, coordinate remaining launch activities, and publish a readiness update to Slack and Notion — all before you flip the switch.

LLM Agent Failure Diagnosis Harness

Systematically diagnose misbehaving LLM agents — classify the failure layer, isolate the root cause, and verify your fix held — using a 7-layer taxonomy built for AI engineers.

LLM Cost Optimization Engine

Cut LLM and agent spend without guessing — turn your usage exports into a costed baseline, a quality-gated model-routing table, a caching and prompt-compression plan, and an A/B measurement contract that proves the savings are real.

LLM Eval System Design Playbook

Design a production eval program for your LLM product feature — measurable quality dimensions, a contamination-guarded golden set, calibrated graders, and CI regression gates with drift monitoring — so quality regressions never reach users on vibes. (Evals the feature itself; for benchmarking coding agents on repo tasks, see the Coding Agent Eval Harness Builder Playbook.)

LLM Incident Response Runbook System

When your LLM feature breaks in production, stop improvising — classify the incident into one of four pre-built classes, assign a defensible severity, walk a class-specific diagnosis tree to the root cause, ship a sequenced containment plan, send the right comms, and close it with a blameless postmortem.

LLM Model Selection Rubric

Pick the right model per workload instead of by habit or hype — extract a workload profile with weighted criteria, score candidates with honest unknowns, run a minimal decisive bake-off, and ship the switch with rollback and revisit triggers.

LLM Observability Setup Playbook

Stand up production observability for your LLM feature — end-to-end traces, evals running on live traffic, cost and quality dashboards, and an alert layer that pages the right owner — with a rollout sequenced by risk and a review cadence that keeps it alive. Works in any chat model; no connected tools required. (For instrumenting one repo with Datadog/GitHub, see the Observability Instrumentation Harness.)

LLM Output Toxicity and Brand Safety Auditor

Gate every customer-facing LLM output before production with a scored rubric across toxicity, brand voice, legal exposure, and geo-cultural sensitivity — giving AI governance, legal, and brand teams a defensible audit trail.

~20 min to use

Other·Advanced

LLM Over-Refusal Audit Rubric

Audit your LLM for over-refusal: classify false rejections across a 7-category taxonomy, score borderline cases with calibrated anchors, map helpfulness-safety tradeoffs, and get a prioritized remediation plan — for ML and product teams in healthcare, legal, and customer support.

Other·Advanced

LLM Output Toxicity & Brand Safety Auditor

Audit LLM outputs for harm, brand misalignment, legal exposure, and geo-cultural risk — four structured rubrics with scored dimensions, deployment verdicts, and prioritised remediation actions for enterprise AI governance teams.

Strategy·Advanced

M&A Sell-Side Data-Room & Diligence Prep Playbook

Get acquisition-ready before buyers do: build a sector-tuned data-room index, red-team your own diligence, draft disclosure schedules, pre-answer buyer questions, and score launch readiness. For founders, CFOs, and corp-dev leads. No connected tools needed.

MCP Tool Wiring Playbook

Design exactly which MCP servers your coding agent needs, scope every permission to least-privilege, and ship a safe-usage guide your team can actually follow.

Sales·Intermediate

MEDDPICC Deal Qualification Rubric

Grade any enterprise deal on evidence, not optimism: a cited MEDDPICC scorecard, gap-closing actions with proof artifacts, a weighted pursue/walk verdict, and the honest forecast category — from raw notes in minutes.

Merge Conflict Resolution Harness

Resolve complex Git merge and rebase conflicts safely by reconstructing each side's intent before touching a single line — designed for Cursor, Claude Code, and Aider agentic workflows.

Research·Advanced

Metric Definition & Governance System

End the "whose number is right" argument: map every conflicting metric definition, write canonical definition contracts down to grain, timezone and late-arriving data, certify and assign owners, retire the duplicates without breaking a dashboard, and catch definitions that silently drift from their code. Works in any chat model, no connected tools.

Model & System Card Writer Pack

Write the model card and system card your governance board and enterprise buyers actually ask for — intended and out-of-scope uses, provenance statements you can substantiate, evaluation results with their caveats attached, a known failure-mode catalog, and the trigger list that keeps the document from going stale. Produces the documentation artifact; your own reviewers decide whether it is sufficient.

Monorepo CI Pipeline Harness Playbook

Design fast, affected-only CI pipelines for any polyglot monorepo — topology, caching, parallelism, and merge gates — without slow full-repo rebuilds.

Monorepo Impact Analysis Agent Pack

Know exactly which packages, services, and tests a change touches before it ships — connected agents read the dependency graph and produce a PR-ready impact report automatically. Requires GitHub connected; Slack optional for team broadcast.

Multi-Agent Architecture Review Board

Put your agent system design in front of a simulated adversarial review board — frame the assumptions, score it on weighted architecture dimensions, run a pre-mortem, and walk away with a board verdict and a blocking-conditions list before you build.

Multi-Agent Workflow Designer Playbook

Turn a messy business process into an implementable multi-agent orchestration spec — agent roles, hand-off contracts, verification points, and failure isolation — instead of one over-scoped agent that silently breaks. Four prompts take you from current-state diagnosis to a sequenced rollout and a governance cadence.

Multi-Repo Change Orchestration Pack

Fan a breaking change across every affected repository — consistent PRs, correct merge order, and a live rollout tracker — without losing a single repo or misaligning a dependency.

NIST CSF 2.0 Gap Assessment & Roadmap System

Run a structured NIST CSF 2.0 maturity assessment in-house: build a Current vs Target profile, score every Function on a consistent tier scale, rank the gaps by risk and effort, and ship a phased remediation roadmap plus a board-ready scorecard — without a five-figure consulting engagement.

Observability Instrumentation Harness

Stop discovering missing telemetry during outages — audit every service for instrumentation gaps, build a consistent tracing and metrics plan, draft the code changes, and generate Datadog dashboard and alert stubs. Requires Datadog and GitHub connected.

Org-Wide Prompt Library Governance System

Run your company's internal prompt and agent library like a design system: inventory and tier every asset, design versioning, ownership, review, and deprecation rules, build the intake and review cadence, then audit for drift so the library stays trusted instead of sprawling into duplicates and orphans.

Outside Counsel Bill Review & Spend Audit Rubric

Catch block billing, admin padding, duplicate entries, and unapproved rate hikes on outside counsel invoices — then ship a firm-ready write-down memo with the dollar amount and rationale per flagged line. Built for GCs, legal ops, and finance partners enforcing billing guidelines they have no time to police.

NeedsDatadogGitHubSSlack

Performance Regression Harness Agent Pack

Catch performance regressions before users feel them — read Datadog baselines, flag latency and throughput degradations, trace them to the responsible deploy, and publish a findings report in one agent session.

Product-Market Fit Measurement System

Measure product-market fit instead of declaring it: read your cohort retention curves for actual flattening, run the disappointment survey with a sample-size gate that refuses unreadable results, separate a pocket of segment fit from whole-product fit, track the leading indicators that move before retention does, and end with one dated decision — double down, narrow, keep searching, or pivot — plus the falsification test that could overturn it. Works in any chat model, no connected tools.

Polyglot Secrets Management Playbook

Eliminate inconsistent secret handling across every service and language in your stack. Four structured prompts inventory where secrets live, design a unified storage and injection standard, produce a phased migration plan, and define rotation and audit practices — giving your platform team a concrete, repeatable security baseline.

Marketing·Advanced

Positioning Workshop Facilitation OS

Facilitate a real positioning workshop end to end instead of generating a slogan: grade your pre-work evidence, map every competitive alternative including doing nothing, break feature claims through attribute-to-value chains, score and tie-break the segment choice when the room deadlocks, then leave with a stress-tested positioning statement and a rollout brief with owners.

Post-AI Process Re-Designer Playbook

Rebuild a process from an empty page now that machines run its middle: find the handoffs, queues, buffers and sign-offs that only ever existed because humans were slow, delete them with a named replacement for what each was quietly doing, re-cut the human roles into charters that say what each one no longer does, and rewrite the turnaround promise and quality bar the new shape can actually hold.

NeedsPostHogSSlackNotion

Experiment Monitoring Agent Pack

Keep every active experiment healthy — monitor PostHog metrics, catch anomalies early, interpret outcome shifts, and push findings to Slack and Notion without manual reporting. Requires PostHog, Slack, and Notion connected to your agent.

Pricing & Packaging Strategy Playbook

Make the highest-leverage decision in SaaS with a system instead of competitor-copying: select a value metric scored on alignment and gaming resistance, architect tiers and fences mapped to willingness to pay, set price points with value anchoring and unit-economics floors, and plan the repricing migration with churn bounds and rollback triggers.

Privacy Policy & ToS Compliance Gap Audit

Audit your own website privacy policy and terms against the live CCPA/CPRA + GDPR disclosure checklist — find missing or vague clauses, score each gap by enforcement severity, and get drafted replacement language you can paste straight in.

Sales·Advanced

Enterprise Procurement & Legal Navigation Playbook

Stop losing quarters inside your buyer's approval process: map every gate from security review to purchase order with its owner and start trigger, run legal, security, and vendor onboarding in parallel instead of in series, prioritize open contract points so counsel sees the ones that matter, and diagnose the exact stall that has stopped the deal. Sales-process preparation for your own counsel — not legal advice.

Productivity·Advanced

NeedsNotionJiraSSlack

Connected Product Requirements Agent Pack

Transform a feature idea into a reviewed, published PRD with linked Jira tickets — your agent reads context from Notion and Jira, drafts the document, breaks it into engineering tasks, and publishes everything in one workflow. Requires Notion and Jira connected.

Production Readiness Review Rubric

Score any service against reliability, observability, scalability, security, and operational readiness before launch — and get a prioritized remediation plan when it falls short.

HR·Advanced

Promotion Packet Builder System

Turn a messy brag doc into a promotion case that survives calibration — a graded evidence inventory, impact quantified against tiers and honest attribution, line-by-line scoring against your own career ladder, a pre-mortem of the ten objections that get voiced in the room, and the written packet plus a one-page sponsor brief.

Agent Prompt-Injection Defense Harness

Harden your coding or automation agent against prompt injection from repo content, issues, code comments, and tool output. Four structured prompts map the attack surface, design input isolation, write enforceable guardrail rules, and generate a red-team test set so you can verify the defenses actually hold.

Prompt Injection Test Corpus Builder

Build a production-ready prompt injection test corpus for your LLM app — systematically covering direct injection, indirect RAG attacks, multi-turn escalation, and tool-call hijacking, with a final coverage map and gap report.

Customer Support·Intermediate

QBR Production System

Produce a renewal-winning QBR in hours, not days: assemble a value-evidence ledger against the customer's own success criteria, narrate ROI with honest gaps and plans, seed expansion only where evidence supports it, and design an executive session that produces commitments instead of nods.

Quarterly Planning OS for Engineering

Run an engineering team's quarterly planning as a system: model real capacity in engineer-weeks after on-call, ramp, holidays, and interrupt load; tier candidate work into Committed / Target / Stretch with confidence bands; map dependencies onto a critical path; and stand up a review cadence plus a mid-quarter replan protocol that says exactly what gets cut first. Works in any chat model — paste your roster, calendar, and work list; no tools required.

RAG Quality Audit Playbook

Stop guessing why your RAG system answers wrong — classify each failure into a nine-family taxonomy, isolate the broken layer with probe queries and a five-lens corpus audit, and ship grounded prompt fixes with verification probes that prove the fix.

NeedsSentryGitHubSSlack

Regression Triage Harness Agent Pack

Pinpoint regression root causes fast — correlate failing CI tests, Sentry error spikes, and suspect commits into a ranked hypothesis list and a filed triage issue, all without leaving your agent.

Regulated Industry Agent Validation Pack

Assemble the validation dossier your compliance function actually asks for before an AI agent goes live in a regulated process: intended-use and impact classification, a control coverage matrix against your own control catalog, a falsifiable test protocol, an adjudicated evidence dossier, and a human-oversight plus revalidation-trigger design. Structures your evidence for internal review — it is not legal advice and not a compliance determination.

NeedsGitHubSSlackNotion

Release Communications Agent Pack

One connected-tool pack for the whole release-comms workflow: collect merged changes from GitHub, classify by audience impact, ship the CHANGELOG.md PR and plain-English release notes from one source, gate the deploy, and publish audience-tuned announcements to Slack and Notion.

NeedsGitHubSSlackJira

Release Train Coordination Harness

Roll up readiness across every service, sequence deploys by dependency, surface rollback paths before they are needed, and publish stakeholder comms — all from one connected harness.

Repo Health Scorecard Rubric

Score any repository on tests, docs, CI, dependencies, and agent-readiness using a weighted rubric — then turn the scorecard into a prioritized, fundable improvement plan engineering leaders can act on immediately.

Repo RAG Index Harness Playbook

Design a production-grade retrieval index over any codebase — chunking strategy, metadata schema, eval queries, and a maintenance loop — so agents and engineers get accurate, cited answers instead of hallucinated snippets.

Research·Intermediate

Research Synthesis Engine

Turn a pile of past studies, surveys, analytics readouts and support exports into one defensible synthesis — every source graded, every claim attributed, every disagreement surfaced instead of averaged away, and a memo whose every sentence traces back to the study that produced it.

Sales·Intermediate

RFP Response Accelerator

Turn an RFP from days of senior time into hours: shred it into atomic requirements with IDs, draft grounded answers that reuse approved language, build the evidence-backed compliance matrix, and weave win themes through the whole response.

Strategy·Advanced

NeedsLinearNotionSSlack

Connected Roadmap Alignment Agent Pack

Keep your roadmap synchronized with business goals and team execution — your agent audits Linear initiatives against strategic objectives, surfaces dependency risks, detects conflicting priorities, and publishes a revised recommendation to Notion with a Slack update. Requires Linear, Notion, and Slack connected.

Roadmap Tradeoff Decision Engine

Turn a contested list of roadmap candidates into a force-ranked, sequenced plan: cost-of-delay and confidence-weighted scoring, an opportunity-cost ledger that names what each approval kills, and a decision memo with a per-stakeholder script for every "not now".

NeedsSSalesforceGmailSSlack

Sales·Advanced

Sales Meeting Follow-Up Agent Pack

Close the loop on every sales meeting before the next one starts: extract decisions, commitments, and objections from raw notes, write them back to the right Salesforce opportunity fields with evidence-grounded accuracy, and send a follow-up that mirrors the buyer's own words — no copy-paste relay across three tools.

NeedsSSalesforceSSlack

Sales Pipeline Hygiene Agent Pack

Run a forecast-honest pipeline audit in one agent session: detect stalled deals from stage-age and activity signals, flag forecast categories the evidence does not support, draft the CRM updates with owner notifications, and hand sales leadership a pipeline they can actually trust.

Security Questionnaire Response Accelerator

Answer inbound enterprise security questionnaires (SIG, CAIQ, bespoke forms) without over-claiming: classify and deduplicate every question, reuse only answers whose evidence is still current, back each affirmative answer with a named artifact, write honest gap answers with compensating controls, and route every binding commitment to a named reviewer.

NeedsSentryJiraSSlack

Production Error Triage Agent Pack

Prioritize Sentry production errors by real user impact, generate actionable remediation plans, and create assigned Jira issues — so your team fixes the errors that matter most instead of triaging noise.

Monolith-to-Services Decomposition Playbook

Identify natural service seams, sequence extractions by risk and value, define clean data boundaries, and design rollback guardrails — ship a decomposition plan that survives contact with production.

Short Thesis Red-Team Rubric

Attack your own short case for the ways it loses money even when the fundamental view is right: scored evidence and a catalyst clock, a borrow and carry model that shows the decline you need just to break even, a weighted crowding and squeeze score, and sizing derived from the adverse move you have to survive. A structuring aid for your own analysis — not investment advice.

SOC 2 Readiness Playbook

Take your startup from zero to audit-ready: a severity-ranked gap register against the Trust Services Criteria, a control map with named owners, an evidence-collection calendar, and an auditor-interaction script — without a $30k consultant.

Spec-to-Code Harness

Turn an ambiguous feature spec into a verified implementation plan and step-by-step guarded agent execution — so your coding agent ships the right feature, not a plausible-looking guess.

NeedsJiraConfluenceSSlack

Productivity·Advanced

Sprint Retrospective Agent Pack

Analyze sprint outcomes, surface recurring blockers, build a concrete improvement plan, and publish the retrospective to Confluence and Slack — so every sprint ends with evidence-based commitments, not just conversation.

Structured Interview Design System

Replace gut-feel interviews with a structured loop that predicts: derive the 5–7 competencies that matter for the role, design behavioral questions with anchored 1–4 rating scales, assemble self-contained interviewer kits, and run a debrief protocol that resists anchoring and halo effects.

Structured-Output Reliability Harness

Stop silent JSON and tool-call drift from breaking your LLM feature in production — design a failure-aware schema, add a self-repair output contract, generate an adversarial regression set, and score every release against a failure budget. Works in any chat model, no connected tools.

Synthetic Data Design Playbook

Design a synthetic dataset for evals or fine-tuning without shipping a contaminated, low-diversity corpus: plan the generation strategy and schema, catch train/eval/test leakage before it poisons your numbers, score coverage and diversity against explicit anchors, and run a go/no-go validation gate before the data is used. Works in any chat model — no connected tools needed.

Tech Debt Portfolio Manager

Turn a vague pile of tech-debt complaints into a fundable quarterly paydown plan: a structured debt register, each item priced as ongoing interest vs one-time principal, an ROI-ordered portfolio that fits your engineering capacity, and an exec-readable business case that survives a CFO's questions. Works in any chat model — paste your context, no tools required.

Technical Design Review Playbook

Turn any architecture proposal into a complete design review package — clarified requirements, scored tradeoffs, mapped failure modes, and a phased execution plan — before a single line of code is written.

Sales·Intermediate

Technical Win Strategy System

Win the technical evaluation without over-claiming: map every evaluator and the pass criteria they are personally judged on, grade what your product genuinely does today, design a POC whose exit criteria are agreed in writing before day 1, position on the criteria this buyer actually weighted, and hand your champion a forwardable technical record that reports the failures as well as the passes.

Research·Advanced

10-K Filing Forensics System

Read filings like a forensic analyst: diff consecutive 10-Ks for the language changes legal let through, sweep footnotes against a red-flag catalog with quoted evidence, run the earnings-quality arithmetic with every calculation shown, and score management's promise-vs-delivery record. Research preparation, not investment advice.

Term Sheet & Cap Table Analyzer

Walk into your first term sheet negotiation knowing exactly what you are signing: every clause decoded into its mechanical effect, the option-pool shuffle and dilution priced with visible arithmetic, exit waterfalls run across good and mediocre sale prices, and a ranked negotiation list plus the question brief for your own attorney. Preparation for counsel — not legal advice.

Test Coverage Gap Harness Pack

Find, rank, and plan tests for the highest-risk coverage gaps across a multi-language codebase — so your next testing sprint targets where failures are most likely, not where adding tests is easiest.

Threat Modeling Workflow System

Threat-model any feature in an afternoon instead of a workshop week: extract a complete data-flow model with trust boundaries, enumerate STRIDE threats as concrete attack narratives, walk the top threats as step-by-step abuse cases with detection gaps, and ship a ticket-ready mitigation plan.

HR·Intermediate

Underperformance Diagnosis Framework

A 4-step system for engineering managers to diagnose an underperforming engineer — separate skill, will, and system causes, stress-test the call for bias, and produce a documented intervention plan with a PIP-readiness gate before any formal step.

Finance·Intermediate

Variance Analysis Narrative Engine

Turn monthly variance tables into driver-attributed, board-ready commentary — price/volume/mix decomposition, evidence-tested one-time vs structural labels, and a forecast-linked review section, every close.

Vendor & Tool Consolidation Audit

Turn raw SSO lists, spend exports, and invoices into a defensible tool-consolidation plan: a normalized inventory, weighted 1-5 overlap and spend scoring, Cut / Consolidate / Renegotiate / Keep verdicts with migration risk grades, and a renewal-aware, owned cut plan. Works in any chat model — no connected tools needed.

Vendor Security Review Rubric

Process SOC 2 reports and security questionnaires into consistent vendor verdicts: extract every control, exception, and CUEC into a normalized register, separate material gaps from cosmetic ones, reason through compensating controls, and ship a tiered approve/conditional/reject recommendation.

~50 min to use

Workflow-to-Agent Migration Kit

Turn an automation you already run — an n8n workflow, a Zapier zap, a cron or Airflow script, an RPA bot — into a defensible migration plan: which steps stay deterministic, which genuinely need an agent, the guardrails each one needs, and the numeric gates that decide promotion or rollback.