The codex-native skill pack for solution engineers

Research accounts, build demos, answer fit-gap, and orchestrate POC handoff. All inside Codex, all one-question conversations. Every claim is cited — or marked unknown. Solutioneer never freehands.

Let Codex bootstrap the skills into ~/.codex/skills, or clone the repo and install manually.

Get Started → GitHub

codex · solutioneer · gpt-5.5

codex

Solutioneer ready · gpt-5.5 · 8 skills loaded

$account-intel-brief outtake.ai

routing → account-intel-brief

grounding via firecrawl…

✓ company snapshot

✓ stack signals · segment · hubspot · snowflake

✓ likely initiatives

✓ discovery questions · cited

returned in chat — nothing saved by default

build a demo for a revops lead at outtake

what you type → what happens

Ask a question in a sentence

Solutioneer routes to the right skill automatically. You can also use explicit $skill syntax when you already know the move.

$account-intel-brief outtake.ai cited company snapshot, stack signals, likely initiatives, question bank

normalize these discovery notes for acme health one fact sheet · slack, gmail, hubspot, granola all welcome

build a demo scenario for a revops lead at outtake narrative, talk track, objections, seeded data, fallback plan

can we integrate with segment and hubspot? support / workaround / gap / unknown matrix with citations

turn this discovery and fit-gap into a poc handoff package poc plan, architecture note, milestones, risks, dependencies

skills

Eight skills across the pre-sales motion

Each one is narrow, chat-native, and composable. The output of one becomes context for the next.

$discovery-context-ingester

Discovery Context Ingester

Normalize messy notes, transcripts, and private-source ingest into a reusable fact sheet.

$account-intel-brief

Account Intel Brief

Pre-call research — company snapshot, stack signals, likely initiatives, discovery questions.

$discovery-question-generator

Discovery Question Generator

Fact sheet → stakeholder-specific questions, signal scorecard, red flags.

$demo-scenario-builder

Demo Scenario Builder

Narrative, talk track, objection pack, deterministic seed data, fallback plan.

$architecture-fit-mapper

Architecture Fit Mapper

Solution architecture map, diagram, blockers, required validations.

$integration-fit-gap-analyzer

Integration Fit Gap Analyzer

Cited support / workaround / gap / unknown matrix against a target stack.

$poc-handoff-orchestrator

POC Handoff Orchestrator

Pre-sales → delivery. POC plan, architecture note, milestones, risks, dependencies.

$security-review-prep

Security Review Prep

Security response brief, control matrix, customer follow-ups, escalation gaps.

how a run works

Every skill follows the same loop

Route, ask the next question, ground, return. No forms. No JSON. No surprise side-effects.

01 · route

Codex picks the skill

Explicit $skill wins; otherwise natural language routes to one of eight. Ambiguous? One routing question first.

02 · ask

One question at a time

Only the next highest-value missing input. No forms, no raw JSON, no re-asking things already in chat.

03 · ground

Cite or mark unknown

Firecrawl wrappers fetch evidence when a key is set. Everything else is labelled unknown or as an assumption.

04 · return

Back in chat

Artifacts land in chat by default. Save or writeback only when you explicitly confirm a destination.

providers

Opt-in integrations

Two providers. Both optional. Both degrade gracefully when keys are missing.

FIRECRAWL_API_KEY

Rules every skill follows

Without configuration. Without exceptions.

Chat-only by default No file writes, no external pushes, unless you ask and confirm.

One question at a time Every skill asks only the next missing input. No JSON forms.

Cited or unknown External claims are either backed by a source or explicitly labelled.

Composable by design Output of one skill is canonical input to the next. No re-research.

Ready to run it?

Clone the repo, install the skills into Codex, and start a conversation.

Get Started → Browse Skills