Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 17 additions & 10 deletions tests/prose/lib/record-action.cjs
Original file line number Diff line number Diff line change
Expand Up @@ -44,14 +44,19 @@ const VIOLATIONS = 'tests/prose/.agent-tool-use.log';
const WORLD = /(^|[\s"'`])(\/[^\s"'`]*\/prose-world-[A-Za-z0-9]+)/;
// These caps exist to protect the asserter's prompt, never to save disk —
// every recorded action is read into it, and a walk makes twenty-odd file
// reads whose bodies are whole skill files. They apply only where the
// content is incidental. A file a walker read back is incidental: no claim
// ever rests on those bytes. A command's output is not — it settles what
// the prose was shown, whether a gate rendered empty, whether a menu had
// entries — so it gets a ceiling generous enough that real evidence is
// never trimmed, bounding only a runaway that would swamp the prompt.
// reads whose bodies are whole skill files.
//
// Only a read is incidental. Nothing is ever claimed about the bytes a
// walker read back out of a file, so those are trimmed hard. Everything a
// walk *produces* is evidence and is kept: what a command returned settles
// whether a gate rendered empty or a menu had entries, and what a write
// put in a file settles every claim about the artifact a phase leaves
// behind. Trimming a write was a read's rule applied to the wrong thing —
// it left a claim about a written file unprovable, because the only copy
// of that content lived in the response being cut.
const MAX_OUTPUT = 400;
const MAX_COMMAND_OUTPUT = 10000;
const MAX_PRODUCED_OUTPUT = 10000;
const WRITE_RESPONSE_TOOLS = new Set(['Bash', 'Write', 'Edit', 'NotebookEdit']);

function read() {
try {
Expand Down Expand Up @@ -213,14 +218,16 @@ function main() {
// instead — so reaching this point is itself the success signal.
parts.push('ok');
parts.push(
responseText(payload.tool_response, tool === 'Bash' ? MAX_COMMAND_OUTPUT : MAX_OUTPUT)
.split(world).join('.'),
responseText(
payload.tool_response,
WRITE_RESPONSE_TOOLS.has(tool) ? MAX_PRODUCED_OUTPUT : MAX_OUTPUT,
).split(world).join('.'),
);
} else if (event === 'PostToolUseFailure') {
parts.push('FAILED');
parts.push(
responseText(payload.tool_response ?? payload.tool_output ?? payload.error,
MAX_COMMAND_OUTPUT).split(world).join('.'),
MAX_PRODUCED_OUTPUT).split(world).join('.'),
);
} else if (stop) {
parts.push(flatten(payload.last_assistant_message, MAX_OUTPUT).split(world).join('.'));
Expand Down
29 changes: 29 additions & 0 deletions tests/scripts/test-prose-record-action.cjs
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,35 @@ describe('prose recorder — tool events', () => {
assert.ok(row.includes('warning: store is stale'));
});

it('keeps what a write put in a file — claims rest on it', () => {
// A read is incidental; a write is the artifact a phase leaves behind.
// Trimming this left a claim about a written file unprovable, because
// the response was the only copy of that content.
const written = `## Symptoms\n${'detail '.repeat(400)}`;
fire({
hook_event_name: 'PostToolUse',
tool_name: 'Edit',
agent_type: 'prose-walker',
tool_input: { file_path: `${world}/.workflows/crash-fix/investigation/crash-fix.md` },
tool_response: { filePath: 'crash-fix.md', newString: written },
});
const [row] = logLines();
assert.ok(!row.includes('[truncated]'), 'the written content survives');
assert.ok(row.includes('detail detail'), 'and is legible in the record');
});

it('keeps a Write the same way', () => {
const written = 'x'.repeat(3000);
fire({
hook_event_name: 'PostToolUse',
tool_name: 'Write',
agent_type: 'prose-walker',
tool_input: { file_path: `${world}/.workflows/notes.md` },
tool_response: { type: 'create', content: written },
});
assert.ok(!logLines()[0].includes('[truncated]'));
});

it('gives a command far more room than a file read', () => {
const long = 'x'.repeat(1800);
fire({
Expand Down