cli
Copilot instructions475 starsChanged 16 days agoMITView on GitHub
This is the Dependabot CLI (dependabot), a Go tool that orchestrates Dependabot update jobs via Docker containers. It does not perform dependency resolution itself — it coordinates three containers: Data flow: CLI parses input → starts proxy + updater containers on isolated Docker networks → updater calls back to the fake API → CLI collects results as YAML. Models use kebab-case YAML tags (yaml:"package-manager") and matching JSON tags. When adding new fields: Each subcommand (update, test, graph) follows the same…
- Reads credentials
- Installs packages
We split this copilot-instructions.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Add omitempty initially to maintain backward compatibility with existing smoke tests
- Very specific: line 36
Call infra.Run(infra.RunParams{...}) with appropriate parameters
≈ “Common” is an estimate: at least 1 in 5 of the 3,669 Copilot instructions 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
Dependabot CLI — Copilot Instructions
Architecture Overview
This is the Dependabot CLI (dependabot), a Go tool that orchestrates Dependabot update jobs via Docker containers. It does not perform dependency resolution itself — it coordinates three containers:
- Proxy (
ghcr.io/dependabot/proxy) — intercepts all updater network traffic, injects credentials without exposing them to the updater, and optionally caches requests. - Updater (
ghcr.io/dependabot/dependabot-updater-<ecosystem>) — runs the actual dependency update logic (from dependabot-core). - Fake API server (
internal/server/api.go) — a local HTTP server that captures updater API calls (create PR, close PR, etc.) for output or test assertion.
Data flow: CLI parses input → starts proxy + updater containers on isolated Docker networks → updater calls back to the fake API → CLI collects results as YAML.
Project Layout
cmd/dependabot/— entrypoint (main), delegates tointernal/cmd/cmd/dependabot/internal/cmd/— Cobra commands:update,test,graph,version, plusroot.gofor shared flagsinternal/infra/— Docker container orchestration:run.go(main flow),updater.go,proxy.go,network.go,config.gointernal/model/— data types for jobs, credentials, smoke tests, API payloads. Shared across all packages.internal/server/— fake API server (api.go) and secure input server (input.go)testdata/— YAML fixtures andscripts/*.txtfor scripttest-based integration tests
Key Conventions
YAML/JSON Model Tags
Models use kebab-case YAML tags (yaml:"package-manager") and matching JSON tags. When adding new fields:
- Add
omitemptyinitially to maintain backward compatibility with existing smoke tests - See the comment block at the top of
internal/model/job.gofor the full add/remove lifecycle
Command Pattern
Each subcommand (update, test, graph) follows the same pattern:
- Define a
NewXCommand() *cobra.Commandconstructor - Call
infra.Run(infra.RunParams{...})with appropriate parameters - Register via
init()withrootCmd.AddCommand() - The
updateandgraphcommands shareextractInput()andprocessInput()fromupdate.go
Credentials Handling
$-prefixed values in YAML input are expanded from environment variables at runtimeLOCAL_GITHUB_ACCESS_TOKENandLOCAL_AZURE_ACCESS_TOKENare auto-injected into credentials when set- Credentials are never passed directly to the updater; they go through the proxy which injects them into outbound requests
checkCredAccess()inrun.goblocks tokens with write access to GitHub API for security
Container Networking
Two Docker bridge networks are created per run (network.go):
- no-internet (internal) — updater can only reach the proxy
- internet — proxy can reach external services
The updater is connected only to no-internet; the proxy bridges both.
Build & Test
# Build
go build ./cmd/dependabot
# Run all unit tests
go test ./...
# Run script-based integration tests (require Docker)
go test ./cmd/dependabot/ -count=1
# Run a specific script test by name pattern
script/e2e <pattern>
# Install from source
go install github.com/dependabot/cli/cmd/dependabot@latestScript Tests (testdata/scripts/*.txt)
These use Go's rsc.io/script framework (see cmd/dependabot/dependabot_test.go). Each .txt file:
- Builds a dummy Docker image inline (via
-- Dockerfile --sections) - Runs
dependabotcommands and asserts on stdout/stderr - Uses
!prefix for expected-failure commands
Test Mocking Pattern
The test command uses a package-level var executeTestJob = infra.Run that tests override to capture RunParams without running Docker (see test_test.go).
Smoke Tests
Smoke tests (model.SmokeTest) define input + expected output for reproducible update jobs:
input:— job definition + credentialsoutput:— array of expected API calls (create_pull_request,update_dependency_list, etc.)- Generate with:
dependabot update <pm> <repo> -o smoke-test.yml - Run with:
dependabot test -f smoke-test.yml --cache ./tmp/cache - The test runner auto-generates
ignore-conditionsto pin dependency versions for reproducibility
Package Manager Ecosystem Mapping
The packageManagerLookup map in internal/infra/run.go maps package manager names (e.g., go_modules) to updater image suffixes (e.g., gomod). When adding ecosystem support, update this map.
Where files are headingNew Copilot instructions files have more rules about branches and pull requests, and fewer about code style.
Of the Copilot instructions files we read, the 2,297 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.
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.

