werkrevisor 0.16.1: Werkator review, review only after WIP, merge by the bot, 100000-token limit #18

Merged

mi merged 13 commits from claude-code/werkator-review-and-automerge into main 2026-10-06 07:36:29 +02:00

Conversation 3 Commits 13 Files Changed 22

+897 -26

claude-code commented 2026-10-05 22:32:24 +02:00

Collaborator

Copy Link

Copy Source

See docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md.

🤖 Generated with Claude Code

See docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-code added 2 commits 2026-10-05 22:32:24 +02:00

PR#18 doc: werkrevisor reviews Werkator, and a ready PR merges itself once it is green ... 7e041ea731

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

werkrevisor 0.16.1: tools/pr-ready, a host config for Werkator and a 100000-token input limit ... ca6317672f

tools/pr-ready removes the WIP prefix and schedules the Gitea merge for when the checks succeed.
config/werkator.toml is a complete host config for mi/werkator.
max_usd_per_review 0.48 raises the derived input limit from 80000 to 100000 tokens.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:18:43 +02:00

werkrevisor 0.16.1: tools/branch-protection sets the approval gate with the token of a repository admin ... 78ac2297e8

It saves the previous rule as a JSON file and keeps its status contexts; --dry-run shows the changes only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:23:18 +02:00

gitignore: backups of tools/branch-protection ... 1c335ecf08

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:23:18 +02:00

ops doc: branch protection of mi/werkator and its backup ... 80a7ef9bf7

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:24:10 +02:00

config/werkator.toml: Werkator's open questions do not block, they record decisions left open ... 830fe4fba3

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:26:32 +02:00

config/werkator.toml: Werkator's open questions use the format of hsadmin-ng since its PR#127 ... a49e143f06

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 06:28:22 +02:00

config/werkator.toml: only dated PR documents count, the template TEAMPLATE.md is none ... 4c17283d7d

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 07:02:03 +02:00

werkrevisor 0.16.1: only a PR changed from WIP is reviewed, and the bot schedules its merge ... 22a0961b46

[gate] require_wip_transition reads the Gitea timeline; a PR opened without WIP gets ERROR, no review, no approval.
[gate] automerge schedules the merge after GREEN, YELLOW and ORANGE; it needs the WIP transition.
tools/pr-ready only removes the WIP prefix now.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

mi added 1 commit 2026-10-06 07:25:02 +02:00

werkrevisor 0.16.1: no tools/pr-ready; the Gitea UI or one PATCH of the title removes WIP ... 555ce9bb66

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude-code changed title from WIP: werkrevisor 0.16.1: Werkator review, tools/pr-ready with automerge, 100000-token limit to werkrevisor 0.16.1: Werkator review, review only after WIP, merge by the bot, 100000-token limit 2026-10-06 07:32:33 +02:00

mi added 1 commit 2026-10-06 07:32:34 +02:00

Merge remote-tracking branch 'origin/main' into claude-code/werkator-review-and-automerge ... 33dc634ecf

# Conflicts:
#	tests/fakes.py
#	werkrevisor/__init__.py
#	werkrevisor/cli.py

ai-reviewer commented 2026-10-06 07:32:41 +02:00

Collaborator

Copy Link

Copy Source

🟡 werkrevisor: YELLOW / HUMAN_REVIEW

Change kind: mixed.

Reasons

• YELLOW risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py)
• YELLOW spec: docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md Scenario#18.05 hard_to_read (0.52 > yellow_above = 0.5)
• YELLOW spec: docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md section 'Follow-up PRs' hard_to_understand (0.51 > yellow_above = 0.5)
• YELLOW 2 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py
• YELLOW no AI review: the PR needs 87692 input tokens (56575 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40

Classifier

Unit	Question	Problem	Confidence	Verdict
docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md Scenario#18.05	hard_to_read	0.52	0.50	YELLOW
docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md section 'Follow-up PRs'	hard_to_understand	0.51	0.35	YELLOW

10 of 12 units passed all questions (1 PR document(s)).
1 answer ignored because of a confidence below [classifier] min_confidence.

_{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 87692, duration 5.4 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10201, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).
Reviewed head `33dc634ecf70f2d46bdb56d38b49f9df7486c373` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

<!-- werkrevisor:33dc634ecf70f2d46bdb56d38b49f9df7486c373 --> <!-- werkrevisor-verdict:YELLOW --> ## 🟡 werkrevisor: YELLOW / HUMAN_REVIEW Change kind: `mixed`. ### Reasons - **YELLOW** risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py) - **YELLOW** spec: docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md Scenario#18.05 hard_to_read (0.52 > yellow_above = 0.5) - **YELLOW** spec: docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md section 'Follow-up PRs' hard_to_understand (0.51 > yellow_above = 0.5) - **YELLOW** 2 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py - **YELLOW** no AI review: the PR needs 87692 input tokens (56575 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40 ### Classifier | Unit | Question | Problem | Confidence | Verdict | |---|---|---|---|---| | docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md Scenario#18.05 | hard_to_read | 0.52 | 0.50 | YELLOW | | docs/prs/2026-10-05-PR#18-werkator-review-and-automerge.md section 'Follow-up PRs' | hard_to_understand | 0.51 | 0.35 | YELLOW | 10 of 12 units passed all questions (1 PR document(s)). 1 answer ignored because of a confidence below [classifier] min_confidence. _{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 87692, duration 5.4 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10201, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).<br> Reviewed head `33dc634ecf70f2d46bdb56d38b49f9df7486c373` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

ai-reviewer requested review from mi 2026-10-06 07:32:41 +02:00

mi added 1 commit 2026-10-06 07:33:33 +02:00

review: PR#18 Scenario#18.05 and Follow-up PRs in plain domain language ... 5ab7f41f30

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

ai-reviewer commented 2026-10-06 07:33:49 +02:00

Collaborator

Copy Link

Copy Source

🟡 werkrevisor: YELLOW / HUMAN_REVIEW

Change kind: mixed.

Reasons

• YELLOW risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py)
• YELLOW 2 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py
• YELLOW no AI review: the PR needs 87738 input tokens (56605 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40

Classifier

12 of 12 units passed all questions (1 PR document(s)).
3 answers ignored because of a confidence below [classifier] min_confidence.

_{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 87738, duration 5.8 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10222, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).
Reviewed head `5ab7f41f307f6020ee5998762367d31e104dd141` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

<!-- werkrevisor:5ab7f41f307f6020ee5998762367d31e104dd141 --> <!-- werkrevisor-verdict:YELLOW --> ## 🟡 werkrevisor: YELLOW / HUMAN_REVIEW Change kind: `mixed`. ### Reasons - **YELLOW** risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py) - **YELLOW** 2 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py - **YELLOW** no AI review: the PR needs 87738 input tokens (56605 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40 ### Classifier 12 of 12 units passed all questions (1 PR document(s)). 3 answers ignored because of a confidence below [classifier] min_confidence. _{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 87738, duration 5.8 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10222, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).<br> Reviewed head `5ab7f41f307f6020ee5998762367d31e104dd141` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

mi added 1 commit 2026-10-06 07:35:22 +02:00

review: the bot schedules a merge only under an approval rule, and a PR set back to WIP is not ready ... 680bca001f

Fixes the two ORANGE findings of the model dry run on PR#18: the base branch must require an approval and status checks,
and only the last WIP change of the title counts.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

ai-reviewer commented 2026-10-06 07:35:30 +02:00

Collaborator

Copy Link

Copy Source

🟡 werkrevisor: YELLOW / HUMAN_REVIEW

Change kind: mixed.

Reasons

• YELLOW risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py)
• YELLOW 3 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py, werkrevisor/readiness.py
• YELLOW no AI review: the PR needs 89750 input tokens (57903 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40

Classifier

12 of 12 units passed all questions (1 PR document(s)).
4 answers ignored because of a confidence below [classifier] min_confidence.

_{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 89750, duration 5.6 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10289, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).
Reviewed head `680bca001f26edcaa46c86717ac69291f6416319` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

<!-- werkrevisor:680bca001f26edcaa46c86717ac69291f6416319 --> <!-- werkrevisor-verdict:YELLOW --> ## 🟡 werkrevisor: YELLOW / HUMAN_REVIEW Change kind: `mixed`. ### Reasons - **YELLOW** risky paths changed: .werkrevisor.toml (built in), werkrevisor/config.py (werkrevisor/config.py), werkrevisor/gitea.py (werkrevisor/gitea.py) - **YELLOW** 3 changed files were given to the AI only as diff, not in full, so it could not check them against the rest of their file (all full files together would exceed [review] max_full_files_chars = 120000): werkrevisor/cli.py, werkrevisor/config.py, werkrevisor/readiness.py - **YELLOW** no AI review: the PR needs 89750 input tokens (57903 counted with o200k_base × [model] token_count_factor = 1.55), more than the limit of 80000 derived from [budget] max_usd_per_review = 0.40 ### Classifier 12 of 12 units passed all questions (1 PR document(s)). 4 answers ignored because of a confidence below [classifier] min_confidence. _{Usage: model `openrouter/anthropic/claude-sonnet-5.5`, LLM skipped (too large), tokens counted before the call 89750, duration 5.6 s, no model call, cost 0 USD, classifier `jev-1.13.0` calls 12, tokens in 10289, cost 0.0004 USD, config: defaults + host + repository (.werkrevisor.toml at base 5ce5cba).<br> Reviewed head `680bca001f26edcaa46c86717ac69291f6416319` against base `5ce5cba6cb45918a8f2a52df40cac7160ad20f75`, werkrevisor 0.16.0.}

mi merged commit 0c61408302 into main 2026-10-06 07:36:29 +02:00

mi deleted branch claude-code/werkator-review-and-automerge 2026-10-06 07:36:30 +02:00

Sign in to join this conversation.

Reviewers

No Reviewers

mi

Labels

No items

No labels

Milestone

No items

No Milestone

Projects

Clear projects

No projects

Assignees

Clear assignees

ai-reviewer claude-code mi

No Assignees

3 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: mi/werkrevisor#18