feat(metrics): investigating-metric-anomalies skill#63132
Open
DanielVisca wants to merge 1 commit into
Open
Conversation
This was referenced Jun 11, 2026
Contributor
Author
This was referenced Jun 11, 2026
Contributor
|
Reviews (1): Last reviewed commit: "feat(metrics): investigating-metric-anom..." | Re-trigger Greptile |
DanielVisca
force-pushed
the
posthog-code/metrics-mcp-characterize
branch
from
4c1ea15 to
53cc0f6
Compare
DanielVisca
force-pushed
the
posthog-code/metrics-investigation-skill
branch
from
a9574c6 to
6174e1a
Compare
DanielVisca
added
the
stamphog
Graphite App
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
DanielVisca
force-pushed
the
posthog-code/metrics-mcp-characterize
branch
from
1752d82 to
06bae3f
Compare
DanielVisca
force-pushed
the
posthog-code/metrics-investigation-skill
branch
from
6174e1a to
6a6d38e
Compare
github-actions
Bot
dismissed
their stale review
New commits pushed (delta classified non_linear_history) — stamphog approval dismissed; re-review running automatically.
The repeatable incident playbook for "metric X looks wrong": pin the metric (metric-names-list), characterize first (one call for magnitude, onset, movers), sharpen hypotheses with targeted query-metrics calls (filters, formula normalization, companion metrics), correlate logs and traces at onset_time, and conclude with evidence. Includes a worked "ingestion lag is rising" example reproduced against a real induced local incident (consumer outage -> backlog drain) where the playbook identified direction, onset within one bucket of the restart, and the exact culprit service via top_movers. How to validate manually: - hogli lint:skills - follow the worked example with the local stack: stop ingestion-logs for a few minutes, restart it, then run the three MCP calls in order Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
DanielVisca
force-pushed
the
posthog-code/metrics-mcp-characterize
branch
from
79d8ba1 to
eebda3f
Compare
DanielVisca
force-pushed
the
posthog-code/metrics-investigation-skill
branch
from
6a6d38e to
8dd4038
Compare
github-actions
Bot
dismissed
their stale review
New commits pushed (delta classified non_linear_history) — stamphog approval dismissed; re-review running automatically.
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
1 participant
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.
Problem
The metrics MCP tools are atomic by design; the investigation knowledge — what order to call them in, how to read the results, when to pivot into logs and traces — needs to live somewhere agents load on demand.
Changes
products/metrics/skills/investigating-metric-anomalies/SKILL.md: the playbook from symptom to evidence — pin the metric (metric-names-list), characterize first (one call: magnitude, onset, movers), sharpen hypotheses with targetedquery-metricscalls (mover drill-down, formula normalization, companion metrics), correlate logs and traces atonset_time, conclude with evidence. Includes a worked "ingestion lag is rising" example and a pitfalls list (counter resets, emitter-died-vs-zero, avg-hides-p95, scrape delay).How did you test this code?
I'm an agent.
hogli lint:skillspasses (78 skills). The worked example is not hypothetical — it's the transcript shape of a real induced incident investigated through the MCP tools (see #63131's test notes).🤖 Agent context
Autonomy: Human-driven (agent-assisted) — directed by @DanielVisca