never-block-a-conversation-on-slow-work

Hand back a handle instead of waiting, tell the person what to expect, and make sure the retry they will attempt anyway does not do the work twice.

Procedure

  1. start start. Goes next to s1
  2. check Measure it at the slow end, not the average. Goes on_fail to s9; next to s2
  3. decision Does it fit inside the budget?. Goes if fits to s7; if default to s3
  4. action Write down the budget and the measurement, RO (read-only). Goes next to __end_success
  5. action Return a handle, immediately, RW (reversible-write). Goes on_fail to s9; next to s4
  6. action Say what will happen and when, RO (read-only). Goes on_fail to s9; next to s5
  7. check A repeat request does not repeat the work. Goes on_fail to s6; next to s8
  8. action Make the submission idempotent, RW (reversible-write). Goes on_fail to s9; retry to s5; next to s8
  9. check Completion reaches the person. Goes on_fail to s9; next to __end_success
  10. escalate Do not ship it synchronous. Terminal
  11. end success. Terminal

Open the graph on its own page — larger, printable, and an address you can send.

Walk it without running it — you answer what your environment would answer, and the path, the capabilities, the approvals and the distance back to an undo come out of the document. Nothing executes; there is no Run button here and there will not be one.

Trust

T2 · attested 2026-10-11 · runbooks.directory

Nobody has attested this record. That is the ordinary case: T2 is what the catalog can check on its own, and T3 needs somebody to run the procedure and sign for it.

  • Link liveness has not been checked: it needs a network and the catalog builds without one. Run pnpm gates --online where there is one.

Steps

  1. s1 check — Measure it at the slow end, not the average
  2. s2 decision — Does it fit inside the budget?
  3. s3 action — Return a handle, immediately (reversible-write)
  4. s4 action — Say what will happen and when (read-only)
  5. s5 check — A repeat request does not repeat the work
  6. s6 action — Make the submission idempotent (reversible-write)
  7. s8 check — Completion reaches the person
  8. s7 action — Write down the budget and the measurement (read-only)
  9. s9 escalate — Do not ship it synchronous

Capabilities

This is a privilege request. Check it against your own allowlist before running.

    Provenance

    Source
    https://github.com/microsoft/ai-agent-runbooks/blob/083f33eec889/02-patterns/Copilot-Studio-and-Foundry-Split-Architecture/Copilot-Studio-and-Foundry-Split-Architecture-Runbook.md
    Fetched
    2026-09-04
    Source last checked
    not since it was captured on 2026-09-04 — recorded as current
    Licence
    MIT
    How it got here
    Rewritten here from the source above, under its licence. The words and the steps are this catalog's; the procedure it describes is not this catalog's invention, and the source is where to check it.
    Highest risk in this procedure
    reversible-write

    Take it with you

    Run it under supervision

    Runs here, on your machine, under supervision: every step is proposed before it happens, an approval gate stops for you, and a failed expectation goes where the procedure says rather than on to the next step. Nothing is sent anywhere. To supervise an agent that is doing the work instead, runbooks hook sits in its tool-call path.

    npx @runbooks/cli run adapted/never-block-a-conversation-on-slow-work@1.0.0 --as "$(whoami)" \
  --input budget=<how long the calling surface will wait before it gives up> \
  --input slow_step=<the work whose duration is in question>