model-tier-context-cap
A deny rule by default: it refuses a leased worker writing past the context its job's model tier caps, and names what to run instead.
What it catches
A leased worker writing past the context its job's model tier caps.
Why
A model tier's bounds in the workspace's declaration can cap the context a worker at that tier reaches (maxContext), because a smaller model past a large context does worse work and resends all of it every turn. When a leased worker's context, as its host last reported it, passes the cap of the tier its row was forked at, every write under the lease is refused, as a lease past its deadline is, and the refusal says to report what it has and stop. Set to advise, the next write is told so once. A tier with no cap, a row with no tier, or a host that reports no context is never judged.
Default and override
By default this rule takes the decision deny. A workspace sets it by name, in its root
magusfile, to deny, advise or off:
magus\guard.builtins({"model-tier-context-cap": "advise"})
A loosening takes effect once it is committed; a tightening applies at once.
Seeing it
A verdict names its rule in brackets, which is how you got here:
deny [model-tier-context-cap]: ...
A workspace that sets this rule to advise sees advise [model-tier-context-cap]: ... instead.
magus describe rule model-tier-context-cap prints the same entry at a terminal, and
magus describe rules lists every rule this workspace enforces.
See also
- All rules - what this workspace enforces, deny first
- The guard - how a verdict is reached and wired
- Setting a built-in rule - how a workspace changes this default