feat: add pre-prompt context size diagnostic logging#8930
Merged
Takhoffman merged 4 commits intoopenclaw:mainfrom Feb 13, 2026
Merged
feat: add pre-prompt context size diagnostic logging#8930Takhoffman merged 4 commits intoopenclaw:mainfrom
Takhoffman merged 4 commits intoopenclaw:mainfrom
Conversation
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
Comment on lines
805
to
813
| // Diagnostic: log context sizes before prompt to help debug early overflow errors. | ||
| { | ||
| const msgCount = activeSession.messages.length; | ||
| const systemLen = systemPromptText?.length ?? 0; | ||
| const promptLen = effectivePrompt.length; | ||
| log.info( | ||
| `[context-diag] pre-prompt: sessionKey=${params.sessionKey ?? params.sessionId} ` + | ||
| `messages=${msgCount} systemPromptChars=${systemLen} promptChars=${promptLen} ` + | ||
| `provider=${params.provider}/${params.modelId} sessionFile=${params.sessionFile}`, |
Contributor
There was a problem hiding this comment.
[P2] sessionFile path logging may leak sensitive local paths.
This info-level diagnostic includes sessionFile=${params.sessionFile}; on multi-user installs or when logs are shared, absolute paths can expose usernames/home directories. Consider logging just path.basename(sessionFile) or a redacted/relative path, or gating this behind a higher verbosity/debug flag.
Prompt To Fix With AI
This is a comment left during a code review.
Path: src/agents/pi-embedded-runner/run/attempt.ts
Line: 805:813
Comment:
[P2] `sessionFile` path logging may leak sensitive local paths.
This `info`-level diagnostic includes `sessionFile=${params.sessionFile}`; on multi-user installs or when logs are shared, absolute paths can expose usernames/home directories. Consider logging just `path.basename(sessionFile)` or a redacted/relative path, or gating this behind a higher verbosity/debug flag.
How can I resolve this? If you propose a fix, please make it concise.
Contributor
Author
|
[P2] Session file path logging: This is consistent with existing logging throughout the codebase — |
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
5 times, most recently
from
6323330 to
3497229
Compare
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
3497229 to
05840c9
Compare
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
6 times, most recently
from
4cc46d8 to
2606b37
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 10, 2026
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
2606b37 to
330c2e8
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 11, 2026
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
2 times, most recently
from
6ccf3f8 to
511bcbb
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 11, 2026
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
511bcbb to
71399c7
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 12, 2026
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
71399c7 to
4cbd8ac
Compare
openclaw-barnacle
bot
added
size: XS
experienced-contributor
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
4cbd8ac to
17b1430
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 13, 2026
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 13, 2026
Glucksberg
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
17b1430 to
3bb69b2
Compare
Glucksberg
added a commit
to Glucksberg/OpenClaw
that referenced
this pull request
Feb 13, 2026
Takhoffman
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
3bb69b2 to
4edc1af
Compare
Log message count, system prompt size, and user prompt size before each session.prompt() call. This helps diagnose early context overflow errors by revealing the actual payload sizes being sent to the model, enabling root cause analysis of unexpected overflows in new or small sessions.
Takhoffman
force-pushed
the
feat/pre-prompt-context-diagnostics
branch
from
4edc1af to
9c8ea69
Compare
Contributor
|
Merged via squash.
Thanks @Glucksberg! |
Contributor
|
This was a good idea. I expanded it a bit so we can get to the bottom of all these tricky compaction bugs. I also added a new log level guard to hide these potential expensive operations behind a less expensive check. Appreciate the PR! |
steipete
pushed a commit
to azade-c/openclaw
that referenced
this pull request
Feb 14, 2026
…thanks @Glucksberg Verified: - pnpm build - pnpm check - pnpm test Co-authored-by: Glucksberg <[email protected]> Co-authored-by: Tak Hoffman <[email protected]>
mverrilli
pushed a commit
to mverrilli/openclaw
that referenced
this pull request
Feb 14, 2026
…thanks @Glucksberg Verified: - pnpm build - pnpm check - pnpm test Co-authored-by: Glucksberg <[email protected]> Co-authored-by: Tak Hoffman <[email protected]>
GwonHyeok
pushed a commit
to learners-superpumped/openclaw
that referenced
this pull request
Feb 15, 2026
…thanks @Glucksberg Verified: - pnpm build - pnpm check - pnpm test Co-authored-by: Glucksberg <[email protected]> Co-authored-by: Tak Hoffman <[email protected]>
cloud-neutral
pushed a commit
to cloud-neutral-toolkit/openclawbot.svc.plus
that referenced
this pull request
Feb 15, 2026
…thanks @Glucksberg Verified: - pnpm build - pnpm check - pnpm test Co-authored-by: Glucksberg <[email protected]> Co-authored-by: Tak Hoffman <[email protected]>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
[context-diag]log before eachsession.prompt()call reporting context sizesProblem
When a session hits context overflow early (e.g., at 15k/200k reported by
/status), there is no visibility into what the actual payload sizes were at the time of the API call. The token counts shown by/statusmay be stale or misleading, making it impossible to diagnose whether the overflow was caused by a large system prompt, accumulated messages, or session contamination.Solution
Add an
info-level log right beforesession.prompt()that reports:messages: number of messages in the active sessionsystemPromptChars: character count of the system prompt (tools, skills, context files, runtime info)promptChars: character count of the user prompt being sentprovider/modelId: which model is being usedsessionFile: path to the session file for cross-referencingThis enables post-mortem analysis of overflow events by correlating the diagnostic log with the overflow error, revealing the actual context composition at the time of failure.
Test plan
pnpm build— passes🤖 Generated with Claude Code
Greptile Overview
Greptile Summary
Adds an
info-level[context-diag]log immediately beforeactiveSession.prompt()in the embedded runner. The log captures message count, system prompt and prompt character lengths, provider/modelId, and the session file path to help diagnose early context overflow issues by correlating the sizes at the time of the API call.Confidence Score: 4/5
session.prompt()and does not alter control flow. Main concern is operational/privacy impact from logging the session file path at info level.