Skip to content

docs(server): reattach recordToolCallAt doc comment - #1690

Merged
Dumbris merged 2 commits into
mainfrom
claude/fix-recordtoolcallat-doc
Oct 20, 2026
Merged

Dumbris merged 2 commits into
mainfrom
claude/fix-recordtoolcallat-doc

Conversation

@Dumbris

@Dumbris Dumbris commented Oct 20, 2026

Copy link
Copy Markdown
Member

#1687 inserted recordOutputBlocked between recordToolCallAt's doc comment and its func line, so the two comments merged and recordToolCallAt lost its doc. This moves the doc comment back. Comment-only change; found in the post-merge review of #1687.

🤖 Generated with Claude Code

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying mcpproxy-docs with  Cloudflare Pages  Cloudflare Pages

Latest commit: 605dbe1
Status: ✅  Deploy successful!
Preview URL: https://2c87955f.mcpproxy-docs.pages.dev
Branch Preview URL: https://claude-fix-recordtoolcallat.mcpproxy-docs.pages.dev

View logs

@codecov-commenter

Copy link
Copy Markdown

⚠️ Please install the 'codecov app svg image' to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@Dumbris
Dumbris enabled auto-merge (squash) October 20, 2026 14:09
@Dumbris
Dumbris merged commit 9dfdb2f into main Oct 20, 2026
50 checks passed
Dumbris added a commit that referenced this pull request Oct 20, 2026
## What

Experiment E5: when an **external** contributor's PR is merged, a bot
posts **one** thank-you comment with an optional, copy-paste share
snippet, and release notes credit external contributors.

- `.github/workflows/contributor-thanks.yml` — `pull_request_target:
closed` (merged only) + `workflow_dispatch` for manual dry runs.
- `scripts/contributors.sh` — `thanks <pr>` and `release <prev-tag>
<ref>`; `scripts/test-contributors.sh` — 26 offline checks.
- `release.yml` — new "Collect contributors" step; a `## Contributors`
section is appended after the LLM-written notes (deterministic, so a
name is never dropped or invented). Existing tooling did not credit
anyone: notes are LLM-summarised commit subjects, git-cliff
`CHANGELOG.md` has no author section.
- `CONTRIBUTING.md` — short "Recognition" note.

Skipped: author `Dumbris`, `algis-dumbris`, any `[bot]`/Bot type, and
association OWNER/MEMBER. Posted as `github-actions[bot]`, never as
Algis. Nothing is posted on the contributor's behalf.

## Comment wording — please approve (rendered for PR #1222, first-time
author)

````text
<!-- contributor-thanks:v1 -->   (hidden marker, used for idempotency)
Thank you @loloDawit! Your PR "fix(test): restore the registry SSRF allow-policy after loopback fixtures" is merged and will ship in the next release.

This is your first merged contribution to MCPProxy. Welcome, and congratulations!      <- only if first-time

If you would like to mention it, here is a snippet you can copy and edit:

```text
My PR to MCPProxy, an open-source MCP proxy for Claude Code, Codex and
Cursor, just got merged:
#1222
Repo: https://github.com/smart-mcp-proxy/mcpproxy-go
```

Sharing genuinely helps a small open-source project get found. It is entirely optional, and we are grateful either way.
````

679 chars with the first-time line (~600 without); ASCII only. "next
release" is used because the release isn't known at merge time. The PR
title is sanitised (one line, `@` neutralised, backticks/brackets/angle
brackets stripped, 120 chars max).

Release-notes section (example, `release v0.70.0 v0.71.1` run
read-only):
```
## Contributors

Thank you to our external contributors for this release: @sb123sb123, @mvanhorn, @AdamMagued, ... @rudycelekli
```

## Rollout — nothing posts until Algis flips it

Every run is a **dry run** (comment printed in the job log, nothing
posted) until the repo variable `CONTRIBUTOR_THANKS_LIVE` is `true`:
`gh variable set CONTRIBUTOR_THANKS_LIVE --body true -R
smart-mcp-proxy/mcpproxy-go`
Manual dry run for any merged PR: Actions > Contributor Thanks > Run
workflow (it only posts if the variable is `true` AND `dry_run` is
unchecked).

## Security (`pull_request_target`)

Checks out only the default branch (`sparse-checkout:
scripts/contributors.sh`, `persist-credentials: false`); never PR head
code. PR data comes via the API, is passed through env/quoted args only,
never interpolated into shell. `permissions: {}` at top, job gets
`contents: read, pull-requests: write, issues: write`. Per-PR
concurrency group + marker check (bot-authored comments only) for
idempotency.

## Testing

- `bash scripts/test-contributors.sh` — 26/26 (eligibility, template,
length < 900, no non-ASCII, hostile titles).
- Read-only dry runs against real PRs: #1222 (first-time=true), #1664
(repeat, first=false), #1690 (maintainer -> skipped). `release v0.70.0
v0.71.1` produced the list above.
- `actionlint` clean on `contributor-thanks.yml`; `release.yml` has the
same pre-existing 53 findings before and after (none from the new step).
`shellcheck -x` clean. `yamllint` not installed. `act` not installed.
**No comment was posted anywhere.**

## Cross-model review (opencode `github-copilot/gpt-6.1-sol`, high)

- Round 1 (the first attempt exited without a verdict — treated as
UNREVIEWED and re-run with a tighter brief): 2 P2 findings, both fixed:
(1) any commenter could suppress the thanks by pasting the marker → only
`github-actions[bot]` comments count; (2) search-index
lag/`incomplete_results` could mis-flag first-time → incomplete =
not-first (residual race for two first-timer PRs merged within index lag
documented and accepted).
- Round 2: **VERDICT: CLEAN** (also ran mocked-API scenarios: empty
comments, human marker, bot marker, later page, incomplete search).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants