infra-context
Skill26 starsChanged 10 days agoMITView on GitHub
How the system under test is built, run and deployed: its architecture and external services, the backend and frontend stacks with their run and test commands, the auth flow, the environments and their URLs, the CI/CD pipelines, and the non-functional budgets it must hold (performance, security, reliability, observability). Load it whenever a task wires the test framework to the stack, touches an environment, a deploy, a pipeline, auth in tests, test IDs, or a performance or security budget, or asks how the product runs or where it is deployed, even when nobody says 'infrastructure' or 'architecture'. Reads its map through `bun run context:map infra-context`. Pure knowledge plus a self-update proposal path: NOT for running a QA stage or editing CI, NOT for business vocabulary (business-domain-context) or data lifecycles (business-data-context).
- Reads credentials
In this repository30 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- acliSkill
- agentic-qa-coreSkill
- agentic-qa-onboardSkill
- bug-screenshot-annotationSkill
- business-api-contextSkill
- business-data-contextSkill
- business-domain-contextSkill
- business-e2e-contextSkill
- evidence-artifactSkill
- exploratory-testingSkill
- framework-developmentSkill
- git-flow-masterSkill
- iql-contextSkill
- jira-administrationSkill
- judgment-daySkill
- orca-orchestrationSkill
- project-contextSkill
- project-discoverySkill
- pr-review-leadSkill
- regression-testingSkill
- session-handoffSkill
- shift-left-testingSkill
- sprint-conductorSkill
- sprint-testingSkill
- test-automationSkill
- test-documentationSkill
- test-framework-adaptationSkill
- xray-cliSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 0 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
DO: read the map through bun run context:map infra-context (or --section backend, --section environments, ...).
- Very specific: line 19
DO: treat a placeholder map as "no infra map".
- Very specific: line 20
DO: take environment URLs from .agents/project.yaml and credentials from .env.
- Very specific: line 22
DO NOT: write anywhere but this skill's own references/.
- Very specific: line 23
DO NOT: copy map content into this SKILL.md.
- Very specific: line 24
Before a step that uses diagrams (redrawing the architecture figure), run the point-of-use check in …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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
infra-context
Kind
contextwith a declared write scope (agentic-qa-core/references/skill-composition-strategy.md§2b). Delivered once by upstream as a placeholder, then owned by the project: the map inside is this project's synthesis. Procedure:agentic-qa-core/references/business-context-maps.md.
Compact Rules
- DO: read the map through
bun run context:map infra-context(or--section backend,--section environments, ...). NEVER readreferences/infra-map.htmlraw: its SVG is most of the bytes and none of the facts. - DO: treat a placeholder map as "no infra map". Say so and hand the user
project-discoveryPhases 2-3; never guess a run command, an environment URL or an auth flow. - DO: take environment URLs from
.agents/project.yamland credentials from.env. The map says which environments exist and how they differ; it never holds a secret or a value that has its own source. - WHEN a session observes something that contradicts a section (a command that no longer runs, a pipeline step, an auth request that changed): PROPOSE the one-section edit with its evidence to the user (or to the conductor when you are a supervised worker), apply it only on approval. Procedure:
references/refresh.md. - DO NOT: write anywhere but this skill's own
references/. No Jira, no.context/, no CI files, no other skill, no product code. - DO NOT: copy map content into this SKILL.md. Judgment goes in
## Rulesorreferences/gotchas.md, dated and measured. - Before a step that uses
diagrams(redrawing the architecture figure), run the point-of-use check inagentic-qa-core/references/preflight-gate.md§8. A table is fine where it says it better.
Read full SKILL.md when: building a briefing for framework adaptation, an environment or CI change, an auth-in-tests decision or a non-functional check; deciding whether a section is stale; or proposing an edit to the map.
What this skill knows
One aspect of the system under test: how it is built and where it runs. Its containers and external services, the stacks and commands of each side, how a user authenticates, which environments exist, how code reaches them, and the budgets that make a slow or fragile build a defect. Loading it changes what the agent KNOWS; the only thing it ever does is propose an edit to its own map.
Sources of truth (cited, never copied)
| Source | What lives there | Role |
references/infra-map.html | the synthesis: overview, architecture, external-services, nfr-<slug>, backend, frontend, environments, ci-cd, discovery-gaps | this skill's map (read via bun run context:map) |
the product repos ({{BACKEND_REPO}}, {{FRONTEND_REPO}}), their manifests, compose files and CI workflows | the real stack, commands and pipelines | wins over the map on any conflict |
.agents/project.yaml environments | environment URLs | the only source for a URL |
.env / .env.example | credentials and their key names | the only source for a secret |
.context/ADR/ | the hard-to-reverse test-architecture decisions | cited from architecture |
project-discovery Phases 2-3 | the generator that CREATEs the map and UPDATEs its stale sections | owns regeneration |
Rules (judgment, dated)
(none yet: each rule carries YYYY-MM-DD · rule · measured: how)
Not here
- Business vocabulary and the business model →
business-domain-context. - Entities, triggers and state machines →
business-data-context. - Endpoint groups and error semantics →
business-api-context. - This framework's own KATA wiring →
test-framework-adaptation, and the code undertests/.
References
references/infra-map.html— the map (generated; read through the reader).references/refresh.md— the self-update procedure and this aspect's staleness signals.references/gotchas.md— measured traps in reading this project's stack.
Refinements
Lessons land as proposals per agentic-qa-core/references/skill-refinement-protocol.md, never as direct edits.
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.

