Characters are defined solely by 5 Areas of Expertise: leader, faceman, fixer, etc. You buy ratings with points. All characters can do all of them to some degree, but whoever's best at each can use a special schtick once per mission. The GM gives a very simple premise/goal, like: "Rescue a kidnapped scientist from Latveria." In the planning phase the players invent all the mission details: where the target is, how they'll get in, what obstacles they'll overcome, etc. They get Mission Points for the amount and quality of stuff they specify. Then there's the execution phase where they attempt to carry out their plan, using their Expertise aided by Mission Points to resolve each point of risk, the result of a roll determining the balance of narrative control between player and GM rather than success/failure. Complications can arise to interfere with the plan. There's a mechanic for agents benefiting by betraying other agents (and the GM as Operations counts as an agent) which sows mistrust among the team. Also, the longer it takes to play through the actual mission (in real time) the harder things get for the agents - there are three suggested ways to handle this.
Liked. Primarily the ideas, of course. It's like a set of notes that can be developed into a full game, maybe twice the length, that addresses aspects of the genre that are often overlooked. Worth a try for a pick-up session with an indie-tolerant group. I'll give it points for an attractive, modern-looking layout too.
Disliked. Well, like many game books it needs another editing pass. There are typos, repetition, unclear bits, terms used before they've been introduced, bits of sections repeated in other sections... (Of course I have a lower threshold for this than most, but I'd say it's not up to to publishing-for-money standard yet.) The traits are defined by astrological codenames like Mars and Pluto, which kind of assumes you have the author's knowledge of symbolism - functional names would be more transparent. The most straightforward reading of the chargen rules says that you have 30 points to spend but can max everything with 5: it relies on the reader working out that costs are cumulative rather than saying so. Dice rolls involve totalling up d6s - could be any number, 3 or 4 for a decent chance of control and probably up to about 8 in practice - which is a bit of a time-taker. The execution phase isn't explained very clearly. It doesn't discuss dire consequences for agents at all - what can happen if things really go wrong for a character?
Afterthought: It's sure as hell the cosmic opposite to the monster that is Spycraft.