Reliability

Follow every inventory page, require completion evidence, then release the planning input. Stop on a timeout, cursor loop or missing evidence.

Claude Code read one page and planned for the whole inventory

A successful MCP read can still be incomplete. Require pagination and snapshot evidence before an agent uses missing items to justify a change.

Worker A holds epoch 41 and pauses. The authority grants epoch 42 to worker B. The resource accepts B and rejects the resumed worker A.

Claude Code restarted. Its old MCP worker kept writing.

A replacement worker does not silence the old one. Fence stale MCP writes at the resource boundary before a resumed job overwrites newer work.

A Claude Code MCP write passes through an acknowledgement gate, durable commit check, and independent read before a durability receipt is issued

Claude Code got a success receipt before the write was durable

An MCP tool can return success after accepting a write but before durable storage. Define acknowledgement levels, verify the committed version, and give Claude Code a durability receipt.

Two parallel Claude Code writes request operation IDs from one atomic identity ledger, which issues unique IDs and blocks collisions before MCP dispatch

Claude Code gave two parallel writes the same operation ID

Two parallel Claude Code tool calls can share an operation ID and corrupt retry evidence. Allocate identity centrally, bind it to intent, and reject collisions before dispatch.

Two Claude Code eval cases are separated by a reset gate that verifies a clean sandbox before the second case starts

Claude Code passed the eval because the previous case cleaned up for it

A Claude Code eval can pass only because an earlier case left the sandbox in a helpful state. Give every case a fresh environment, a state fingerprint, and a reset receipt.

A timed-out Claude Code MCP request leaves a paid remote job running while a guarded retry looks up the operation and attaches to the existing job

Claude Code timed out, retried, and paid twice

A timed-out MCP call may still be running and charging your account. Give paid tool operations a stable ID, a spend ceiling, and a retry receipt before Claude Code tries again.

A Claude Code patch enters an independent evidence review before the agent explanation is revealed in a second pass

Do not let Claude Code explain its patch to the evaluator first

A polished Claude Code handoff can anchor the reviewer before the evidence is checked. Give an independent evaluator the task contract, patch, and raw test results first, then reveal the agent’s explanation.

A Claude Code trace with a missing MCP tool event is blocked by a completeness gate before replay and review

Make Claude Code prove its trace is complete before replay

A clean replay can be dangerously reassuring when the original trace dropped the tool call that changed production. Add sequence checks, expected-event accounting, and a completeness receipt before review.

Claude Code compacts its conversation while an external control-state checkpoint preserves authority, denied actions, pending effects, and evidence bindings

Keep Claude Code's safety state outside its context window

Context compaction can preserve the task while losing the decision that blocked a dangerous tool call. Store authority, denials, and unresolved effects in a control checkpoint outside the transcript.

Three individually valid Claude Code MCP reads are rejected because their source versions form a mixed production snapshot

Reject a Claude Code plan built from a mixed snapshot

Claude Code can read three correct facts and still build a plan for a system state that never existed. Record a coherent read set and reject evidence that crosses a consistency boundary.