admin
+ + +-
+
++ +
LLM or Deterministic Tool? A Decision Framework for Recurring AI Work
Choose code or a model from the task’s acceptance criteria, uncertainty and verification needs—not novelty or a universal cost claim. Read the article →
-
+
++ +
Shared AI Operating Rules: Reduce Repeated Decisions Without Promising Token Savings
Shared operating rules can make recurring safeguards easier to inspect; paired tests are still needed before claiming token or quality gains. Read the article →
-
+
++ +
What Is Token Optimisation? A Measurement-First Definition
A practical definition: improve recurring AI workflow efficiency while preserving the agreed result, quality checks and verification. Read the article →
-
+
++ +
Context Budgets Are an Allocation, Not a Target to Fill
A method for allocating relevant instructions, evidence and output reserve without filling a model context window by default. Read the article →
-
+
++ +
Subscription vs API: Compare Cost per Verified Job, Not a Token Price Guess
A method for comparing subscription and API costs using the same verified-job window without inventing subscription token rates. Read the article →
-
+
++ +
WAIT_AUTH: Separate Authentication Recovery from Resource Waiting
Treat credentials and provider capacity as separate wait states, preserve the checkpoint and measure human recovery rather than guessing. Read the article →
-
+
++ +
A Plan Change Coincided with OpenCode Auth Failures: What the Logs Actually Show
Redacted logs show four auth-invalidated records across two sessions; timing is measured, but the plan-change cause and resumed jobs remain unknown. Read the article →
-
+
++ +
WAIT_RESOURCE: What a Provider-Free Resume Test Can—and Cannot—Show
A fake-adapter wait test can exercise checkpoint and retry invariants; it cannot establish live provider recovery or accepted output. Read the article →
-
+
++ +
Surviving AI Usage Limits: Keep Recovery Outside the Model Turn
Persist job state outside a constrained model request and test resource-wait invariants without mistaking simulation for production evidence. Read the article →
-
+
++ +
How to Detect a Stalled AI Agent Without Killing a Legitimately Long Job
Separate process liveness, provider activity, semantic progress and acceptance so quiet legitimate work is not mistaken for a stall. Read the article →