SeniorOS
AGENTS.md2 starsChanged 15 days agoApache-2.0View on GitHub
SeniorOS is a vendor-portable engineering harness for coding agents. Keep the core host-neutral and keep permanent context small. Read only the files relevant to the task. Do not map the entire repository for a trivial change. When adding or materially changing a skill: For runtime or skill changes, run: For documentation-only changes, run the relevant validators if the docs affect skill/eval structure. Keep commits atomic. Do not rewrite shared branch history, push, publish packages, deploy, or create releases unless explicitly…
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 repository7 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- senior-code-reviewSkill
- senior-debuggingSkill
- senior-engineeringSkill
- senior-finishSkill
- senior-project-intelligenceSkill
- senior-skill-authoringSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
14 rules. About 0 are common, 9 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Read only the files relevant to the task.
- Distinctive: line 19
Never claim a command, test, build, or check passed unless it was actually run and its result observed.
- Distinctive: line 20
Prefer minimal coherent changes over broad refactors.
- Distinctive: line 21
Do not add a dependency when Node built-ins are sufficient without explaining why.
- Very specific: line 22
Do not overwrite existing installed skills unless the user explicitly requests it or --force is part of the requested operation.
- Distinctive: line 23
Keep vendor-specific behavior out of core skills when an adapter can own it.
- Very specific: line 24
Avoid secrets, credentials, personal data, or raw conversation logs in .senior/.
- Distinctive: line 29
When adding or materially changing a skill:
- Very specific: line 31
keep SKILL.md focused: under 500 lines of body and under 20,000 bytes (npm run validate:skills enforces both).
- Very specific: line 33
put deep reference material in one-level-deep references/ files.
- Distinctive: line 34
put deterministic repeated mechanics in scripts.
- Distinctive: line 36
add or update at least three eval cases: trigger, non-trigger/low-risk, and failure/edge behavior.
- Very specific: line 37
run npm run validate:skills and npm run validate:evals.
- Distinctive: line 51
Keep commits atomic. Do not rewrite shared branch history, push, publish packages, deploy, or create releases unless explicitly requested.
≈ “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
SeniorOS repository instructions
Purpose
SeniorOS is a vendor-portable engineering harness for coding agents. Keep the core host-neutral and keep permanent context small.
Source of truth
- Architecture boundaries:
ARCHITECTURE.md - Roadmap:
ROADMAP.md - Skill authoring contract:
skills/senior-skill-authoring/SKILL.md - Eval contract:
evals/README.md - CLI/runtime:
src/
Read only the files relevant to the task. Do not map the entire repository for a trivial change.
Engineering rules
- Never claim a command, test, build, or check passed unless it was actually run and its result observed.
- Prefer minimal coherent changes over broad refactors.
- Do not add a dependency when Node built-ins are sufficient without explaining why.
- Do not overwrite existing installed skills unless the user explicitly requests it or
--forceis part of the requested operation. - Keep vendor-specific behavior out of core skills when an adapter can own it.
- Avoid secrets, credentials, personal data, or raw conversation logs in
.senior/. - High-risk changes must include proportionate verification and rollback/recovery thinking.
Skill changes
When adding or materially changing a skill:
- keep
SKILL.mdfocused: under 500 lines of body and under 20,000 bytes (npm run validate:skillsenforces both); - make routing metadata describe both what and when;
- put deep reference material in one-level-deep
references/files; - put deterministic repeated mechanics in scripts;
- define explicit evidence/exit criteria;
- add or update at least three eval cases: trigger, non-trigger/low-risk, and failure/edge behavior;
- run
npm run validate:skillsandnpm run validate:evals.
Verification
For runtime or skill changes, run:
npm run checkFor documentation-only changes, run the relevant validators if the docs affect skill/eval structure.
Git hygiene
Keep commits atomic. Do not rewrite shared branch history, push, publish packages, deploy, or create releases unless explicitly requested.
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.