Which suite did that check join?

A client says the build passed, or says the commit never joined a suite. Tonight, three proofs, in a repository the agency or the client can open. Name the suite the check joined: the head sha, the status, and the conclusion. If the response stored no suite id, write that. If the conclusion is null, write null. Show a different suite whose sha, or whose status, is not this one. Show another client's repository that does not list this suite. If any of the three needs the person who installed the app, you have not named the suite. You have a green mark and a guess.

This is one check suite. It is not the check a commit reported. Which check did that commit report? is a check-run id, a name, and a conclusion. A check-run id can sit beside a suite and still not be the suite. It is not the protection a branch set. Which protection did that branch set? is a review count and a lock. A review count is not a sha. It is not the clock a test froze. Which clock did that test freeze? is a clock id and a frozen time. It is not the detail a response omitted. Which detail did that response omit? is a reason and a token cap. A suite is none of those.

None of what follows is a finding that this suite is the right one for the client, and none of it is legal advice. The fields below are one host's current description of a check suite. The create example prints a request and two status codes, and it does not print a suite id. The figures below that are a paper about other people's workflow minutes, not the conclusion on this commit. A share of someone else's minutes is not the suite the check joined tonight.

GitHub's create example sends one field. The head sha is d6fde92930d4715a2b49857d24b940956b26d2d3. 40 characters. The page says a 200 means the suite already exists, and a 201 means the suite was created. 200 is not 201. The same request body is what both codes answer. The page does not print a response id, a status, or a conclusion under either code. A status code is not a suite. The status names the schema lists are queued, in_progress, completed, waiting, requested, pending, and null. Six names plus null is 7. The page does not print 7. queued is 6 characters. in_progress is 11. 6 is not 11. The conclusion names are success, failure, neutral, cancelled, skipped, timed_out, action_required, startup_failure, stale, and null. Nine names plus null is 10. 10 is not 7. cancelled is 9 characters. timed_out is also 9, and it is not the same string. action_required is 15. startup_failure is also 15, and it is not the same string. The page says a rerequest resets status to queued and clears the conclusion. queued is not success, and a cleared conclusion is not a conclusion of success. The note at the top says a GitHub App usually receives one check_suite event per commit sha, even if that sha is pushed to more than one branch. 1 is not 2. One event is not a second suite. The API version named on the page is 2026-03-10. 10 characters, the same number as the conclusion count, and not a suite id (GitHub, REST API endpoints for check suites).

The preferences example on the same page is a different request, and it is not a suite. It sends auto_trigger_checks as an array of one object. app_id is 4. setting is false. The page says the default setting is true. false is not true. 4 is not 40, the length of the sha. 1 object in the array is not a suite id of 1, because the response schema does not print a suite id at all. The path is check-suites/preferences. A preference that turns automatic creation off does not name the suite a push already created, and a preference that leaves it on does not name it either. You must have admin on the repository to set the preference. Admin is not a conclusion. Do not paste app_id 4 onto the sha because both examples live on the suite page.

The minute totals people quote when they want a cost for this build did not open this suite. Islem Bouzenia and Michael Pradel, in the ICSE 2024 paper, study GitHub Actions workflows. The abstract says the platform offers 2,000 minutes of computation free to public repositories each month, that an average paid-tier repository costs $504 a year, and that testing and building consume 91.2 percent of used resources, triggered by pull requests 50.7 percent, pushes 30.9 percent, and scheduled workflows 15.5 percent. 50.7 plus 30.9 plus 15.5 is 97.1. 97.1 is more than the 90.0 percent the introduction uses for the most common events, and 97.1 is not 100. The page does not print 97.1. 91.2 is not 50.7. Caching is adopted by 32.9 percent of paid-tier repositories. A timeout is used by 14.0 percent of paid-tier repositories, affects 4.3 percent of runs, and reduces VM time by 8.1 percent. 32.9 is not 14.0. 3.4 percent, the caching reduction they also print, is not 8.1. 2,000 divided by 60 is 33.333 hours. The introduction calls the same allowance 33 hours. 33 times 60 is 1,980 minutes. 1,980 is not 2,000. The gap is 20 minutes. The page does not print 20 (Bouzenia and Pradel, Resource Usage and Optimization Opportunities in Workflows of GitHub Actions).

The same paper's totals do not collapse into the monthly average. 952 repositories use 23.7 million minutes, with an average of 1,614 monthly minutes. 23.7 million divided by 952 is 24,894.96, and that divided by 1,614 is 15.42. The page does not print 15.42. A total is not a month. 75.1 percent of repositories stay at or under 2,000 minutes, and 24.9 percent go over. 75.1 plus 24.9 is 100. That pair adds. The paid-tier average is 5,914 minutes a month, and the free-tier average is 87. 0.249 times 5,914 plus 0.751 times 87 is 1,537.923. 1,537.923 is not 1,614. The page does not print 1,537.923. 5,914 minus 2,000 is 3,914. The page does not print 3,914. 2,000 divided by 87 is 22.99, not an integer the page prints. 9.9 million runs are 75.4 percent of runs and 96.4 percent of VM time. 75.4 is not 96.4. 9.9 divided by 0.754 is 13.13. The page does not print 13.13 as the run count. The failure sentence says 10.0 percent to 17.4 percent fail. The table's failure row also prints 30.9. 30.9 is outside 10.0 to 17.4, and 30.9 is the same written figure as the push share, which is not a failure rate. 78.7 plus 17.4 is 96.1. 88.9 plus 10.0 is 98.9. 66.3 plus 30.9 is 97.2. 81.1 plus 18.0 is 99.1. None of those is 100. Running previously failed jobs first saves 1.1 percent of VM time across all runs and 31.6 percent on failed runs. 31.6 divided by 1.1 is 28.73. The page does not print 28.73. None of these is the sha, and none of them is a conclusion (Bouzenia and Pradel, DOI 10.1145/3597503.3623303).

Do the three proofs on one real suite, in the repository that actually holds it. First, the binding. Copy the suite id if the response has one, the head sha, the status, and the conclusion. If the only thing you were shown is a 200 or a 201, write the code, and write that the example page did not store an id under it. If the conclusion is null, write null. Do not infer success because the client remembers a green mark. Do not copy app_id 4 into the suite id. Second, the other suite. A different sha, even with status queued, is not this one. A rerequest that clears the conclusion is not a second suite unless the id changed. A paper's percentage is not that second suite. Third, the other repository. A list that belongs to another client does not contain this sha. If the only list is one shared organization and you cannot say whose repository it is, write that. A tenant you cannot name is the finding.

Three moves feel like the proof and are a second problem. Rerunning the suite because the conclusion was embarrassing, and then describing the cleared field as a pass. The page says a rerequest resets status to queued and clears the conclusion. Cleared is not success. Quoting 91.2 or $504 because they look precise, while the suite you were asked about is one object, and the paper that printed them had already failed to make its trigger shares add to 100 and had already failed to make its tier averages weight back to 1,614. The yearly $504 is the paper's average paid-tier estimate. It is not an invoice on this repository. Asking a model which suite the check joined. The model will pick a plausible status. It will not open the object. A check you have not kept is which check did that commit report?. Opening that check does not name this suite.

The path is ordinary when an operator can open one client's suites without opening the next client's suites as the way to prove it. Gigbase is an agency workspace for projects, clients, and contracts, distinct from the next client's repository. It is not a check-suite ledger. Nothing in that sentence is a headcount, a duration, or a result figure. If the only record is a screenshot of a green mark, you do not have this proof. You have a slide.

There is a version of this call where taking the work is the wrong purchase. The work is not sold under the agency's name. The client asked to meet the builders. The room only has a slide. The buyer only wants a template. Or the buyer wants a legal opinion about a status check. A white label software development partner who needs the client's repository in order to see the suite is describing the engagement. One who can only quote 91.2 percent is not. 91.2 is not your suite, and the sha on the example is not your suite, and a 200 is not an id.

When the client is real and the repository cannot name the suite, the missing object is specific. A record, in a repository the agency or the client can open, of the sha, the status, and the conclusion, or a written line that no suite id was stored, plus a different suite that is not that one, plus another client's list that does not hold it. If the first slice is a fixed capability, an AI integration sprint is the shape where the acceptance sentence can include the three proofs. If you do not yet know which repository stores the check, a two-week AI readiness assessment maps the systems. The findings are yours. They are not a promise that 2,000 free minutes will appear on the client's invoice. If what you are buying is the product around the repository, the offer is an AI product development company.

Before you tell the client which suite the check joined, do the three proofs and write the sha. Name the sha, the status, and the conclusion, or write that the response stored no suite id. Show a different suite that is not that one, even if the sha was pushed to a second branch. Show another repository that does not contain it. Do not paste success onto a cleared conclusion, and do not paste 30.9 from the push row onto the failure row because both cells print 30.9. The binding is the suite you can defend.