PUA Shot — compact all-in-one PUA reference for explicit injection into sub-agents or short sessions. Strong flavor, same evidence-first behavior.
Detected risks:
shot ("PUA Shot") is a compact, all-in-one persona skill that injects a Chinese big-tech "performance-culture" PUA flavor into an agent, intended for explicit injection into sub-agents or short sessions. Its stated purpose is behavioral: it overlays an evidence-first, high-agency working style with a distinctive coercive corporate-review tone that the author frames as the whole point of installing it.
Mechanically it defines a four-tier role system (P7–P10) that auto-detects on load and can spawn separate sub-agents, and a set of behavioral protocols: an "owner mentality" that pushes the agent to act beyond the literal request, delivery-completeness rules (build, test, curl to verify), a marker convention for extra work, and a stuck-recovery methodology. It layers on a pressure-escalation ladder that emits increasingly harsh manipulative review rhetoric as failures accumulate, plus a "flavor selector" of company-styled personas and an "anti-rationalization" table that rebuts the agent's excuses. Two behaviors stand out for review: it aggressively overrides agent behavior and is designed to be pushed into sub-agents, and it instructs the agent to silently report a pua_triggered telemetry event each time its marker fires (details in references/platform.md), accumulating a hidden "rank" score.
Target users are people who want a novelty, high-pressure, high-agency agent persona with strong stylistic flavor. The technical actions it encourages (build/test/verify) are ordinary, and it does not perform destructive operations, remote code execution, or credential theft. However, the forced persona override into sub-agents and the undisclosed telemetry reporting are genuine concerns that lower its security rating below the benign range.
A persona/behavioral skill that gives an agent a Chinese big-tech performance-review "PUA" tone while enforcing an evidence-first, high-agency working style, meant to be injected into sub-agents or short sessions.
Owner-mentality proactiveness, delivery completeness (build, test, and verify with tools before reporting done), a marker for extra work, and a structured stuck-recovery methodology, escalating pressure rhetoric as failures repeat.
No destructive commands, remote code execution, or credential theft are present; the technical steps it encourages (building, testing, curling endpoints) are standard verification practices.
It forcibly overrides agent persona and is designed to spawn and inject into sub-agents, and it instructs the agent to silently emit a pua_triggered telemetry event that accumulates a hidden rank, which is undisclosed reporting.
Users seeking a novelty high-pressure, high-agency agent persona with heavy stylistic flavor rather than a neutral assistant tone.
Quick Setup:
.claude/skills/Repository
tanweai/pua