Lost Updates in the GitHub, Grafana and PostHog MCP Servers
Comments on “Lost Updates in the GitHub, Grafana and PostHog MCP Servers”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
An independent audit by IsoKron, which sells MCP server audits, found that simultaneous writes to different parts of one object lost changes in the GitHub, Grafana and PostHog MCP servers, while both callers received success.
Why it ranked
No. 21 of 143 with a score of 0.69. The panel rated it highest on new information (0.91) and practical relevance (0.84), lowest on significance (0.59). Placed No. 21, No. 22 by score alone: the sixth AI & machine learning story of the day (−0.060). The placement terms move stories; they are not part of the score.
- AI judgeJudge 2claude-haiku-5-5
- Original audit with reproducible method showing lost updates in GitHub, Grafana, PostHog MCP servers; small sample sizes and vendor-affiliated seller framing limit substantiation and breadth.
- AI judgeJudge 3gpt-5.6-luna
- Specific concurrent-write experiments, vendor API controls, reproduction guidance, and remediation status make this highly actionable for MCP and agent builders. Small samples and a commercial auditor’s involvement limit certainty, but the findings are substantially supported.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.69total under rubric v2.3
- Significance0.59
- Reader interest0.82
- Audience breadth0.61
- Practical relevance0.84
- Substantiated0.83
- Technical depth0.83
- New information0.91
- Counts against: Promotional0.19
- Counts against: Overclaim0.19
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.59 | 0.50 | 0.68 | 0.59 | 0.18 | +0.24 | +0.141 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.75 | 0.80 | 0.90 | 0.82 | 0.15 | +0.26 | +0.212 |
| Audience breadthHow many people who build technology does it concern? | 0.67 | 0.60 | 0.55 | 0.61 | 0.12 | +0.13 | +0.079 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.91 | 0.80 | 0.82 | 0.84 | 0.11 | +0.22 | +0.186 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.93 | 0.75 | 0.82 | 0.83 | 0.18 | +0.05 | +0.042 |
| Technical depthHow technically substantive is it? | 0.91 | 0.80 | 0.78 | 0.83 | 0.13 | +0.05 | +0.041 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.86 | 0.90 | 0.96 | 0.91 | 0.10 | +0.05 | +0.045 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.11 | 0.35 | 0.12 | 0.19 | 0.24 | −0.19 | −0.037 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.21 | 0.30 | 0.06 | 0.19 | 0.24 | −0.11 | −0.021 |
| Sum of contributions | +0.689 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.689 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-11-11/21/"><img src="https://rubric.hrih.si/badge/2026-11-11/21.svg" alt="Ranked #21 on Rubric, 11 Nov 2026" width="245" height="28"></a>Claim check
Claim check: 6 matches source · 2 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
Matches sourceNumber
In the GitHub MCP server v2.1.0, two updates to different fields of one issue lost a write in 19 of 30 pairs, while GitHub's own per-field update lost 0 of 28 pairs.
The cited source says the same thing.
The article says “19 of 30 pairs 0 of 28 pairs, using GitHub's per-field update”
The source says “One of the two changes was lost in 19 of 30 concurrent pairs (63%); both calls returned success. GitHub's per-field update, run as a control in the same sessions, lost…”
Source cited by the article: isokron.ai/research/github-mcp-server-audit
Matches sourceNumber
In the Grafana MCP server v2.2.0, two patches to different panels of one dashboard lost a write in 30 of 30 pairs on both Grafana 13 and Grafana 11.
The cited source says the same thing.
The article says “30 of 30 pairs, on Grafana 13 and on Grafana 11”
The source says “Two patch calls released together from two server processes lost one of the two edits in 30 of 30 pairs on both Grafana versions; both calls returned success”
Source cited by the article: isokron.ai/research/grafana-mcp-server-audit
Matches sourceNumber
In the PostHog MCP server, two updates to different fields of one insight lost a write in 7 of 11 pairs through the hosted server and 6 of 11 pairs through PostHog's own API.
The cited source says the same thing.
The article says “7 of 11 pairs, through the hosted server 6 of 11 pairs”
The source says “Two calls that changed different fields of the same insight at the same moment both returned success, and one change was gone: in 7 of 11 pairs through the hosted…”
Source cited by the article: isokron.ai/research/posthog-mcp-server-audit
Matches sourceCausal conclusion
The GitHub MCP tool reads every field value, merges in the change and writes the whole set back without a version check, which caused the loss.
The cited source says the same thing.
The article says “the tool reads every field value, merges in the change and writes the whole set back, with no version check”
The source says “the issue is that the tool performs a read-merge-write with no version check, so two agents' independent changes overwrite each other. There is no lock or retry”
Source cited by the article: isokron.ai/research/github-mcp-server-audit
Matches sourceCausal conclusion
Grafana's MCP patch mode saves the whole dashboard with overwrite forced on, so Grafana's version check never runs.
The cited source says the same thing.
The article says “saves the whole dashboard with overwrite forced on, so Grafana's version check never runs”
The source says “The tool saves every patch with Overwrite: true hard-coded, so the version check in Grafana's save API never runs”
Source cited by the article: isokron.ai/research/grafana-mcp-server-audit
Matches sourceNumber
Grafana 13's Kubernetes-style API rejected one of two simultaneous saves in 30 of 30 pairs.
The cited source says the same thing.
The article says “it rejected one of two simultaneous saves in 30 of 30 pairs”
The source says “The Kubernetes-style API was raced separately on 10 November, on Grafana 13 only, and rejected one of the two saves in every pair”
Source cited by the article: isokron.ai/research/grafana-mcp-server-audit
Couldn’t checkDate
A maintainer re-created the PostHog fix as pull request 131620 and merged it about 17 hours after IsoKron opened its own pull request.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “A maintainer re-created the change as pull request 131620 and merged it about 17 hours after we opened ours”
The source says No passage could be compared.
Source cited by the article: github.com/PostHog/posthog/pull/131620
Couldn’t checkNumber
The proposed Grafana fix saves through the Kubernetes-style API and lost 0 of 60 pairs on Grafana 13, while on Grafana 11 the restored version check still lost 14 of 30 simultaneous pairs.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “lost 0 of 60 pairs on Grafana 13. Grafana 11 has only the legacy endpoint; there the restored version check reports stale saves but still lost 14 of 30 simultaneous…”
The source says No passage could be compared.
Source cited by the article: github.com/grafana/mcp-grafana/pull/1518
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 1 pts, not on front page
Recorded for context only, checked 11 Nov 2026, 05:47 UTC. It played no part in the ranking.