Skip to content

Improve context management of process tools #290

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
wonderwhy-er merged 5 commits into from
Dec 11, 2025
Merged

Improve context management of process tools #290

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
ac33eca
Improve context managment of process, truncate outputs and allow offs…
wonderwhy-er Dec 10, 2025
9ff639c
fix: length parameter now caps tail reads in process output pagination
wonderwhy-er Dec 10, 2025
3355b68
fix: tail reads now properly respect length parameter and show accura…
wonderwhy-er Dec 10, 2025
c739f54
fix: getOutputSinceSnapshot now checks completedSessions as fallback
wonderwhy-er Dec 10, 2025
450d1ab
Cleanup
wonderwhy-er Dec 10, 2025
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
11 changes: 10 additions & 1 deletion package-lock.json
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

42 changes: 22 additions & 20 deletions src/server.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -780,35 +780,37 @@ server.setRequestHandler(ListToolsRequestSchema, async () => {
{
name: "read_process_output",
description: `
Read output from a running process with intelligent completion detection.
Read output from a running process with file-like pagination support.

Automatically detects when process is ready for more input instead of timing out.
Supports partial output reading with offset and length parameters (like read_file):
- 'offset' (start line, default: 0)
* offset=0: Read NEW output since last read (default, like old behavior)

@codeant-ai codeant-ai Bot Dec 10, 2025

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggestion: The documentation for the process output tool incorrectly states that offset=0 always reads "NEW output since last read", but in the implementation this only holds for active sessions and not for completed sessions where repeated offset=0 calls re-read from the beginning, which can lead callers to misuse the API and repeatedly stream the same output. [logic error]

Severity Level: Minor ⚠️

Suggested change
* offset=0: Read NEW output since last read (default, like old behavior)
* offset=0: For running processes, read NEW output since last read (default behavior)
Why it matters? ⭐

This is a documentation correctness suggestion, not a code bug fix. The current comment in the tool description implies a guarantee that may not hold for completed sessions (many process-read implementations will return from start for finished processes or when there is no last-read cursor). Clarifying the text to "For running processes, read NEW output since last read (default behavior)" avoids misleading callers and reduces incorrect assumptions about API semantics. It's a safe, useful clarification that improves developer UX and prevents subtle misuse.

Prompt for AI Agent 🤖 
This is a comment left during a code review.

**Path:** src/server.ts
**Line:** 787:787
**Comment:**
	*Logic Error: The documentation for the process output tool incorrectly states that `offset=0` always reads "NEW output since last read", but in the implementation this only holds for active sessions and not for completed sessions where repeated `offset=0` calls re-read from the beginning, which can lead callers to misuse the API and repeatedly stream the same output.

Validate the correctness of the flagged issue. If correct, How can I resolve this? If you propose a fix, implement it and please make it concise.

* Positive: Read from absolute line position
* Negative: Read last N lines from end (tail behavior)
- 'length' (max lines to read, default: configurable via 'fileReadLineLimit' setting)

SMART FEATURES:
- Early exit when REPL shows prompt (>>>, >, etc.)
- Detects process completion vs still running
- Prevents hanging on interactive prompts
- Clear status messages about process state
Examples:
- offset: 0, length: 100 → First 100 NEW lines since last read
- offset: 0 → All new lines (respects config limit)
- offset: 500, length: 50 → Lines 500-549 (absolute position)
- offset: -20 → Last 20 lines (tail)
- offset: -100, length: 50 → Last 50 lines (length caps tail)

OUTPUT PROTECTION:
- Uses same fileReadLineLimit as read_file (default: 1000 lines)
- Returns status like: [Reading 100 lines from line 0 (total: 5000 lines, 4900 remaining)]
- Prevents context overflow from verbose processes

REPL USAGE:
- Stops immediately when REPL prompt detected
- Shows clear status: waiting for input vs finished
- Shorter timeouts needed due to smart detection
- Works with Python, Node.js, R, Julia, etc.
SMART FEATURES:
- For offset=0, waits up to timeout_ms for new output to arrive
- Detects REPL prompts and process completion
- Shows process state (waiting for input, finished, etc.)

DETECTION STATES:
Process waiting for input (ready for interact_with_process)
Process finished execution
Timeout reached (may still be running)

PERFORMANCE DEBUGGING (verbose_timing parameter):
Set verbose_timing: true to get detailed timing information including:
- Exit reason (early_exit_quick_pattern, early_exit_periodic_check, process_finished, timeout)
- Total duration and time to first output
- Complete timeline of all output events with timestamps
- Which detection mechanism triggered early exit
Use this to identify when timeouts could be reduced or detection patterns improved.

${CMD_PREFIX_DESCRIPTION}`,
inputSchema: zodToJsonSchema(ReadProcessOutputArgsSchema),
annotations: {
Expand Down
Loading