test(prose): discovery, walked to the point where work first exists

[leeovery]

Summary

Discovery is the phase every work type begins at, and it had no coverage at all. This walks it from workflow-start on a genuinely empty project — no work units, no inbox, not even a project manifest, since none exists until the first work unit is created — through the opener, the shape-and-confirm gate and name resolution, stopping the moment the work unit is on disk.

That stop is the durability boundary, not an arbitrary point. Everything before the work-type commit is ephemeral however long the shaping ran; everything after it persists. The case asserts exactly that: nothing on disk until one engine transaction creates the work unit, installs the cache-staged session log, and commits.

Three firsts:

• First case that creates a work unit rather than reading one someone else built.
• First live user of world: "claims" — the manifest description and the session log's prose are the model's to write, so the world is judged against stated claims rather than byte-compared.
• Only the second case where engine_before_write is not vacuous, since the walk stages the session log under .workflows itself.

The deterministic spine is ordering: manifest exists must run before workunit create — a collision check that runs after the thing it checks for is no check at all. Plus calls_exclude on the epic machinery (no map, no topics, no session close), since single-phase work must not touch any of it.

The name is supplied by the scripted user rather than accepting the suggestion, which fixes every path in the expected world instead of leaving it to what the model would derive from the description.

Test plan

Passed on first run, on claude-sonnet-5:

PATH: 5/5 steps passed
WORLD: PASS
PASS  engine_before_write — every workflow write followed an engine call
PASS  calls_include  — manifest exists search-filters, workunit create search-filters feature
PASS  calls_exclude  — ran none of: discovery-map, discovery-session close, topic start
PASS  calls_in_order — manifest exists search-filters → workunit create search-filters feature
MARKERS: none

Deterministic gate 76/76; snapshot rebuilds byte-identical.

🤖 Generated with Claude Code

Stack

1. docs(design): prose-tests programme design log #544
2. feat(prose-tests): the framework — cases, worlds, runner, skill #545
3. test(prose): feature happy-path corpus — five worlds, seven cases #546
4. test(prose): bugfix corpus — the investigation-centric surfaces #548
5. test: retry recursive teardown removals — kill a class of phantom failures #549
6. fix(entry-skills): close the handoff fences — six files render their arms wrong #550
7. docs: a contributing page for working on the system #551
8. fix(entry-skills): every handoff arm says to invoke the skill #552
9. fix(implementation): environment setup belongs to the setup reference alone #553
10. fix(prose-tests): the asserter is told which substitutions were armed #554
11. feat(prose-tests): the mid-flow substitution, and a world only prose can describe #555
12. test(prose): claims assert consequences, not what was displayed #556
13. feat(prose-tests): record everything the agents do, results included #557
14. fix(discussion-entry): the handoff reports the source it actually had #558
15. fix(prose-tests): the stop hook records, and names the model that walked #559
16. fix(prose-tests): command output was never actually recorded #560
17. feat(prose-tests): judge the walk as told, not the summary returned #561
18. feat(prose-tests): decide in code what an agent should not be deciding #562
19. test(prose): a case starts where a session starts #563
20. feat(prose-tests): walk on Sonnet, judge on Opus, escalate a failure #564
21. test(prose): give the eight read-only cases something that can fail #565
22. test(prose): only walks that can be observed, and checks that survive the trip #566
23. fix(prose-tests): the verdict names only the model the record names #567
24. test(prose): discovery, walked to the point where work first exists #568 👈 current
25. fix(prose-tests): the asserter judges which of prose or walker was at fault #569
26. docs(conventions): a step whose reference routes every exit still signposts #570
27. test(prose): discovery's epic arm, to the same durability boundary #571