BitSquare/Codeman
2
0 Fork 0
mirror of https://github.com/Ark0N/Codeman.git synced 2026-11-15 02:49:33 +01:00
Code Issues Packages Projects Releases Wiki Activity

fix(story-mode): whose idea a failure was, new repositories everywhere, honest counts (review round 10)

Browse Source 
From the person who asked and a fact-checker, reading fresh pages:
- "as planned" for failing first runs the prompt asked for ("it should
  fail because mode is missing", "run them to watch them fail"): those
  read "The first run fails on purpose, as you asked".
- A repository Claude started was told only on a save panel; the first
  save inside a loop or a test run says it too ("Claude starts a git
  repository and saves the work in 4 commits along the way.").
- `git push && git describe --tags; echo "exit=$?"` exits 0, so a
  failure after the push went unsaid; what it printed after the push
  (`fatal:`, `exit=128`) now counts. The commit subject stays on that
  caption too.
- Tool-call counts no longer include second opinions.
- "a layout check" is "a formatting check"; "Turn over" (a game word)
  is "Not done".
- "the last check finds nothing" told nothing: it now says what came
  back empty ("the last one finds a git setting that is not set, which is
  no error").

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Codeman maintainer committed 2026-10-18 17:16:03 +02:00
1 parent 05dc6fbd95
commit 30b907860d
15 files changed
+161 -78
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/story-mode.ts
+66 -34
View File
@@ -2011,7 +2011,9 @@ export function buildStoryTurns(entries: StoryEntry[]): StoryTurn[] {
}
const NOT_TOOL_CALLS = new Set<StoryIntent>(['prompt', 'say', 'interrupt', 'compact', 'notice']);
const countToolCalls = (events: StoryEvent[]) => events.filter((e) => !NOT_TOOL_CALLS.has(e.intent)).length;
/** The turn's own tool calls: a second opinion (the advisor) is no step of the work, so it is not counted. */
const countToolCalls = (events: StoryEvent[]) =>
events.filter((e) => !NOT_TOOL_CALLS.has(e.intent) && e.tool !== 'Advisor').length;
export function summarizeTurn(turn: StoryTurn): StoryTurnSummary {
return {
@@ -2332,10 +2334,11 @@ function printableLine(line: string): string | null {
return clip(text.replace(/[\w.+-]+@[\w-]+(\.[\w-]+)+/g, '…@…').replace(/[A-Za-z0-9+/=_-]{32,}/g, '…'), 90);
}
/** What a check other than the tests is, by its tool, for a reader: "a layout check", "a style check". */
/** What a check other than the tests is, by its tool, for a reader: "a formatting check", "a style check". */
function checkName(e: StoryEvent | undefined): string {
const cmd = e?.label ?? '';
if (/\b(ruff format|black|prettier|gofmt|rustfmt|isort|autopep8|biome format)\b/.test(cmd)) return 'a layout check';
if (/\b(ruff format|black|prettier|gofmt|rustfmt|isort|autopep8|biome format)\b/.test(cmd))
return 'a formatting check';
if (/\b(ruff|eslint|flake8|pylint|clippy|shellcheck|golangci-lint|stylelint|biome)\b/.test(cmd))
return 'a style check';
if (/\b(tsc|mypy|pyright|vue-tsc)\b/.test(cmd)) return 'a type check';
@@ -2360,13 +2363,36 @@ const emptyAnswer = (e: StoryEvent): boolean => {
);
};
/** The first save of a folder Claude made a repository (`git init`, not `--bare`) with nothing committed before. */
function startsRepoAt(indices: number[], events: StoryEvent[]): boolean {
const earlier = events.slice(0, Math.min(...indices));
return (
!earlier.some((e) => (e.commits ?? 0) > 0) &&
[...earlier, ...indices.map((i) => events[i])].some(
(e) => e.tool === 'Bash' && /\bgit init\b(?![^;&|\n]*\s--bare\b)/.test(e.label)
)
);
}
/** What the last lookup of an empty answer came back without, in a reader's words. */
function emptyAnswerWhat(e: StoryEvent): string {
const step = EMPTY_ANSWER_STEP.exec(e.label.trim())?.[0].replace(/^[;&|\s]+/, '') ?? '';
return /^git config/.test(step)
? 'the last one finds a git setting that is not set'
: /^(which|command -v)/.test(step)
? 'the last one finds a program that is not installed'
: /^(test|\[)/.test(step)
? 'the last test it makes comes out false'
: 'the last search finds no match';
}
/** One or two files by name ("calc.py and test_calc.py"), more by count ("3 files"). */
const fileWords = (files: string[], noun = 'file'): string =>
files.length && files.length <= 2 ? files.map((f) => basename(f)).join(' and ') : plural(files.length, noun);
/** You asked for a failing run first (test first), or Claude said the failure was expected. */
const PLANNED_IN_PROMPT =
/\b(test[- ]first|tdd|fail(s|ing)? (first|on purpose)|failing tests? first|on purpose|red first|before (the )?(code|module|function|implementation) exists)\b/i;
/\b(test[- ]first|tdd|fail(s|ing)? (first|on purpose)|failing tests? first|on purpose|red first|before (the )?(code|module|function|implementation) exists|should fail because|(watch|see) (them|it|the tests?) fail|must (error|fail) (at|on|because|while))\b/i;
const EXPECTED_BY_CLAUDE =
/\b(as expected|as planned|as intended|expected (failure|to fail|red)|deliberately fail\w*|on purpose|test[- ]first|the red state)\b/i;
function plannedFailure(turn: StoryTurn, events: StoryEvent[]): false | 'asked' | 'planned' {
@@ -2587,7 +2613,7 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
const testFails = fails - lintFails;
// A failure you asked for (test first), or that Claude said it expected, is the plan working.
const planned = passed(checks[0]) === false ? plannedFailure(turn, events) : false;
// The check that failed while the tests passed, by name: "a layout check", not "another check".
// The check that failed while the tests passed, by name: "a formatting check", not "another check".
const otherCheck = checkName(
failed.find((e) => !!e.tests && e.tests.passed > 0 && !e.tests.failed && !e.tests.errors)
);
@@ -2617,7 +2643,10 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
: `Fix and test, ${plural(n, 'round')}.`;
// Commits made inside the loop (a test one-liner that ends in one) are part of its story.
const saved = evs.reduce((k, e) => k + (e.commits ?? 0), 0);
if (saved) story += ` Claude saves the work${saved > 1 ? ` in ${saved} commits` : ''} along the way.`;
if (saved)
story += ` Claude ${startsRepoAt(draft.events, events) ? 'starts a git repository and ' : ''}saves the work${
saved > 1 ? ` in ${saved} commits` : ''
} along the way.`;
director = `${plural(checks.length, 'check')} · ${plural(edits.length, 'edit')} · ${fmtCounts(panel.added, panel.removed)} · last: ${
lastCheck ? outcome(lastCheck) : 'running'
}`;
@@ -2684,7 +2713,9 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
if (saved) {
director += ` · ${plural(saved, 'commit')}`;
if (ok)
story = `${story.slice(0, -1)}, and Claude saves the work${checks.some((e) => e.ships === 'push') ? ' and sends it up' : ''}.`;
story = `${story.slice(0, -1)}, and Claude ${startsRepoAt(draft.events, events) ? 'starts a git repository and ' : ''}saves the work${
checks.some((e) => e.ships === 'push') ? ' and sends it up' : ''
}.`;
}
break;
}
@@ -2703,7 +2734,7 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
story = evs.every((e) => e.running)
? `Claude starts ${evs.length === 1 ? 'a command' : `${evs.length} commands`} in the background.`
: errors && errors === evs.length && evs.every(emptyAnswer)
? `Claude checks ${evs.length === 1 ? 'a few things' : 'several things'}; the last check finds nothing.`
? `Claude looks a few things up; ${emptyAnswerWhat(evs[evs.length - 1])}, which is no error.`
: errors && errors === evs.length
? `Claude tries ${evs.length === 1 ? 'a command' : `${evs.length} commands`}, and ${
evs.some((e) => /\bblocked\b|PreToolUse|denied/i.test(e.output ?? ''))
@@ -2731,6 +2762,10 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
(ships === 'push'
? /\[new branch\]|\S+\.\.\S+\s+\S+\s+->|\bset up to track\b|Everything up-to-date/.test(out)
: /^\[[^\]\n]+ [0-9a-f]{7,}\]/m.test(out));
// `git push && git describe; echo "exit=$?"` exits 0, but printed the failure after the push.
const pushedAt = ships === 'push' ? out.search(/\[new branch\]|\S+\.\.\S+\s+\S+\s+->|Everything up-to-date/) : -1;
const hiddenFail =
last?.ok === true && pushedAt >= 0 && /^(fatal|error): |\bexit[= ][1-9]\d*\b/m.test(out.slice(pushedAt));
if (savedAnyway) panel.ok = true;
const commits = evs.reduce((n, e) => n + (e.commits ?? 0), 0);
const inCommits = commits > 1 ? ` in ${commits} commits` : '';
@@ -2767,33 +2802,30 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
)
);
const stepAfter = looksOnly ? 'a check after that fails' : 'a step after that fails';
story = savedAnyway
? uploadOnly
? `Claude ${uploads('the work')}, but ${stepAfter}.`
: ships === 'push'
? `Claude saves the work${inCommits} and ${uploads('it')}, but ${stepAfter}.`
: `Claude saves the work${inCommits}, but ${stepAfter}.`
: panel.ok === false
? ships === 'pr'
? 'Claude tries to open a pull request, and the command fails.'
story =
savedAnyway || hiddenFail
? uploadOnly
? `Claude ${uploads('the work')}, but ${stepAfter}.`
: ships === 'push'
? 'Claude tries to upload the work, and the command fails.'
: 'Claude tries to save the work, and the command fails.'
: ships === 'pr'
? 'Claude opens a pull request.'
: uploadOnly
? `Claude ${uploads('the work')}.`
? `Claude saves the work${inCommits}${said} and ${uploads('it')}, but ${stepAfter}.`
: `Claude saves the work${inCommits}${said}, but ${stepAfter}.`
: panel.ok === false
? ships === 'pr'
? 'Claude tries to open a pull request, and the command fails.'
: ships === 'push'
? localTarget && said
? `Claude saves the work${said}, then ${uploads('it')}.`
: `Claude saves the work${inCommits} and ${uploads('it')}${said}.`
: `Claude saves the work${inCommits}${said}.`;
? 'Claude tries to upload the work, and the command fails.'
: 'Claude tries to save the work, and the command fails.'
: ships === 'pr'
? 'Claude opens a pull request.'
: uploadOnly
? `Claude ${uploads('the work')}.`
: ships === 'push'
? localTarget && said
? `Claude saves the work${said}, then ${uploads('it')}.`
: `Claude saves the work${inCommits} and ${uploads('it')}${said}.`
: `Claude saves the work${inCommits}${said}.`;
const wrote = unique(edits.map((d) => d.file));
const firstAt = Math.min(...draft.events);
const earlier = events.slice(0, firstAt);
const startsRepo =
!earlier.some((e) => (e.commits ?? 0) > 0) &&
[...earlier, ...evs].some((e) => e.tool === 'Bash' && /\bgit init\b(?![^;&|\n]*\s--bare\b)/.test(e.label));
const startsRepo = startsRepoAt(draft.events, events);
const lead = [
startsRepo ? 'starts a git repository' : '',
wrote.length ? `writes ${fileWords(wrote)}` : '',
@@ -2804,7 +2836,7 @@ function decorate(draft: DraftPanel, events: StoryEvent[], turn: StoryTurn, inde
story =
lead.length === 2
? `Claude ${lead.join(', ')}, and ${rest}`
: `Claude ${lead[0]}${rest.split('"')[0].includes(' and ') ? ',' : ' and'} ${rest}`;
: `Claude ${lead[0]}${rest.replace(/"[^"]*"/g, '').includes(' and ') ? ',' : ' and'} ${rest}`;
}
director = `${evs.map(shortCommand).slice(0, 2).join(' · ')}${errors ? ` · ${plural(errors, 'error')}` : ''}${
edits.length ? ` · writes ${plural(unique(edits.map((d) => d.file)).length, 'file')}` : ''
@@ -2975,7 +3007,7 @@ export function buildStoryPage(turn: StoryTurn): StoryPage {
? { finishedLater: events.filter((e) => e.finishedLater && e.intent !== 'prompt').length }
: {}),
};
const calls = events.filter((e) => !NOT_TOOL_CALLS.has(e.intent));
const calls = events.filter((e) => !NOT_TOOL_CALLS.has(e.intent) && e.tool !== 'Advisor');
// A check that found nothing answered its question: it did not fail to work.
if (calls.length && calls.every((e) => e.ok === false && !emptyAnswer(e))) base.noneWorked = true;
if (turn.endedAt) base.endedAt = turn.endedAt;
src/web/public/story-mode.js
+1 -1
View File
@@ -2524,7 +2524,7 @@ Object.assign(CodemanApp.prototype, {
: page.superseded
? 'To be continued…'
: page.noneWorked
? 'Turn over'
? 'Not done'
: page.helpersRunning || page.finishedLater || page.events.some((e) => e.running && !e.helper)
? // Work was still out when the turn ended (and maybe still is): not a clean finish.
'Done for now'