qubership-ai-agent-telemetry
AGENTS.md1 starsChanged 24 days agoApache-2.0View on GitHub
- Write every committed file in English, including documentation, code comments, commit messages, and identifiers. - Treat README.md and docs/ as maintained documentation. Files under docs/superpowers/ are historical snapshots; never update them to describe later changes. - For a design choice that materially changes scope or behavior, ask the user. Present the recommended option first and expect it to be challenged. - Never run apm compile: it rewrites the repository-root agent instruction files. Use apm install to refresh committed harness…
- Deletes or force-pushes
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository24 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .cursor/rules/apm-authoring.mdcCursor rule
- .cursor/rules/change-description-authoring.mdcCursor rule
- .cursor/rules/docs-page-authoring.mdcCursor rule
- .cursor/rules/english-developer-style.mdcCursor rule
- .cursor/rules/markdown-line-length-120.mdcCursor rule
- .cursor/rules/qubership-repo-essentials.mdcCursor rule
- .cursor/rules/qubership-workflow-hub-usage.mdcCursor rule
- apm-authoringSkill
- change-description-authoringSkill
- docs-page-authoringSkill
- english-developer-styleSkill
- markdown-line-length-120Skill
- qubership-actions-guideSkill
- qubership-templates-guideSkill
- qubership-workflow-conventionsSkill
- apm-authoringSkill
- change-description-authoringSkill
- docs-page-authoringSkill
- english-developer-styleSkill
- markdown-line-length-120Skill
- qubership-actions-guideSkill
- qubership-templates-guideSkill
- qubership-workflow-conventionsSkill
The file
We split this AGENTS.md into the rules it gives an agent
10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 5
Write every committed file in English, including documentation, code comments, commit messages, and identifiers.
- Very specific: line 6
Treat README.md and docs/ as maintained documentation.
- Very specific: line 10
Never run apm compile: it rewrites the repository-root agent instruction files.
- Distinctive: line 12
Never run git clean -xdf: it deletes ignored and unrelated untracked work.
- Very specific: line 17
Before starting work, committing, pushing, or updating a pull request, run git fetch origin and then git merge-base --is-ancestor …
- Distinctive: line 19
If another worktree or user coordination prevents a safe update, stop and ask the user.
- Distinctive: line 22
Keep the pull-request description aligned with the current scope and verification steps after each scope change.
- Distinctive: line 23
After each push, inspect checks for the exact head SHA.
- Distinctive: line 25
Use Conventional Commits; the pull-request workflows enforce the convention.
- Very specific: line 31
Before each push, derive the validation set from git diff --name-only origin/main...HEAD, not from the task's apparent subsystem.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Repository agent instructions
Non-obvious invariants
- Write every committed file in English, including documentation, code comments, commit messages, and identifiers.
- Treat
README.mdanddocs/as maintained documentation. Files underdocs/superpowers/are historical snapshots; never update them to describe later changes. - For a design choice that materially changes scope or behavior, ask the user. Present the recommended option first and expect it to be challenged.
- Never run
apm compile: it rewrites the repository-root agent instruction files. Useapm installto refresh committed harness assets. - Never run
git clean -xdf: it deletes ignored and unrelated untracked work. Preview withgit clean -xdn, then remove only the intended generated paths.
Delivery workflow
- Before starting work, committing, pushing, or updating a pull request, run
git fetch originand thengit merge-base --is-ancestor origin/main HEAD. Fast-forwardmainor rebase another branch when the check fails. If another worktree or user coordination prevents a safe update, stop and ask the user. - Send every change, including documentation, through a ready-for-review pull request, not a draft. Merge only after one approval, all review threads are resolved, and the required checks pass for the current head.
- Keep the pull-request description aligned with the current scope and verification steps after each scope change.
- After each push, inspect checks for the exact head SHA. Report pending, failed, or cancelled checks separately and do not describe CI as complete until every expected check is terminal.
- Use Conventional Commits; the pull-request workflows enforce the convention.
- Release only from
mainthrough theReleaseworkflow with avMAJOR.MINOR.PATCHinput. The workflow creates the tag; never push release tags manually.
Validation
- Before each push, derive the validation set from
git diff --name-only origin/main...HEAD, not from the task's apparent subsystem. Match every changed path against the workflow trigger filters and internalpaths-filterrules, then run every locally available command from each activated workflow. Record platform-only checks as unavailable. - For Go changes, run
make testandgo vet ./...;make testincludes the race detector. - The root
Makefiledoes not cover backend or installer behavior. For those changes, follow the commands in.github/workflows/telemetry-backend-tests.yaml,.github/workflows/installer-tests.yaml, and.github/workflows/bootstrap-tests.yaml.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

