Catalog/Messy Middle Navigation

Messy Middle Navigation

Make sure to use this skill whenever an AI feature is partially working but not reliable enough for full autonomy, deciding how much human oversight to maintain alongside AI automation, calibrating the automation level for a human-in-the-loop workflow, shipping an AI feature that works most of the time but fails unpredictably, or defining graduation criteria for increasing AI autonomy — even if the request is about 'AI deployment' or 'automation' rather than explicitly about the messy middle.

Lenny's Podcastengineeringadvisorheuristic
Download .skillInstall instructions

Install

Run the one-liner to fetch and unpack the skill into your Claude Code skills directory:

curl -sL https://skill.broker/skills/michael-truell/messy-middle-navigation.skill -o /tmp/messy-middle-navigation.skill && \ unzip -o /tmp/messy-middle-navigation.skill -d ~/.claude/skills/messy-middle-navigation/

Manual install

Download the .skill file above, then unzip it into ~/.claude/skills/messy-middle-navigation/.

Related skills

Ai Deployment Flywheel Free
Make sure to use this skill whenever deploying an AI agent or model into production and the team expects it to work perfectly on day one, evaluating an AI vendor that claims 'one-click deployment,' or designing the data-feedback loop that makes an AI system improve post-deployment — even if the request is about 'AI deployment' or 'agent design' rather than explicitly about flywheels.
engineeringadvisorheuristicAishwarya Naresh Reganti Kiriti Badam
Trust Battery Check Free
Make sure to use this skill whenever a working relationship seems strained, decisions are being second-guessed without clear rationale, collaboration between two parties is degrading, a new partnership needs a baseline trust assessment, or delegation levels need calibration — even if the request is about 'collaboration' or 'working relationships' rather than explicitly about trust.
engineeringadvisorheuristicBrandon Chu
Artifact Driven Debate Free
Make sure to use this skill whenever a product debate has lasted more than 2 meetings, participants argue from different mental models with no shared reference, a design or architecture decision needs convergence, the team is going in circles with abstract arguments, or a decision requires comparing approaches side by side — even if the request is about 'decision-making' or 'alignment' rather than explicitly about artifacts.
engineeringadvisorproceduralDylan Field