Drop reason from TechreadException

[werk24]

The owner decided on 2026-09-30 to drop reason completely (core-reader#2453). A READ_INCOMPLETE warning tells the customer that a read is incomplete, never why.

History of the field:

• It was added in Refuse a callback read too large for its request by name; model a completeness warning on COMPLETED #587 and has been on PyPI since 2.8.0.
• Only core-reader#2450 ever filled it, and only on core-reader's test branch, for 54 minutes. core-reader#2452 replaced that code.
• No production core-reader release has ever sent it.

The change:

• werk24/models/v2/internal.py: remove the field and its docstring lines. The READ_INCOMPLETE docstring now says the warning never says why. ask_type stays.
• werk24/cli/commands/techread.py: stop printing , reason <x>. That line came in with Clearer first-run experience: size check, key handling, readable errors, JSON CLI, notebooks #591, after #2453 was written. Without this change, werk24 techread would crash on any message carrying an exception.
• Tests:
  • The existing tests no longer use the field.
  • A new test pins that the model has no reason.
  • Another new test pins that a message still carrying "reason" parses, with the key ignored and never dumped. The model keeps pydantic's default of ignoring unknown keys.

Compatibility. Nothing changes on the wire, because the server never sent the field. It is a removal from a released public model, though. Code on 2.8.0 that reads exception.reason directly would get AttributeError after upgrading; getattr(exc, "reason", None) keeps working. The docs-v2 changelog PR announces it.

Merge order:

1. core-reader#2511 first. core-reader installs werk24 from main unpinned, and two of its tests read the field.
2. Then this PR.
3. werkflow pins werk24 at 80db94a, so it is not affected until it bumps. Its own PR removes the field there.

Tests.

• .venv/bin/python -m pytest -q -p no:cacheprovider (Python 3.13, from requirements.txt, tests/requirements.txt and -e .): 999 passed, 5 skipped. The 5 skipped are the credentialed live tests.
• The two changed test files: 28 passed, against 26 on main. The difference is the 2 new tests.

Questions for the owner

One question. It does not block the merge.

1. Release this as 2.8.1?
   • A removal from a public model would usually wait for a major version. But the field has been out for two days and has never carried a value, so a patch release removes it before anyone builds on it.
   • Without a release, PyPI users keep 2.8.0 with the empty field. core-reader and docs-v2 already track main.
   • Recommendation: yes, 2.8.1, with a release note pointing at the docs-v2 changelog entry.

🤖 Generated with Claude Code

https://claude.ai/code/session_01H5VZH5w3jzcmxijpeamRaB

---

Generated by Claude Code

[Copilot]

Copilot review overview

🟢 Approval recommended

The focused removal is consistently applied, preserves legacy wire compatibility, and has appropriate regression coverage.

Review effort: Balanced
Findings: None

What changed in this PR

Removes the unused reason field from TechreadException while preserving compatibility with legacy payloads.

Changes:

• Removes the model field and CLI rendering.
• Updates documentation and tests.
• Confirms legacy reason keys are ignored during parsing.

File	Description
werk24/​models/​v2/​internal.py	Removes reason from the model and documentation.
werk24/​cli/​commands/​techread.py	Stops displaying the removed field.
tests/​test_read_incomplete_warning.py	Tests field removal and legacy payload handling.
tests/​test_cli_techread_command.py	Updates expected CLI exception output.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.