Skip to content

GravionLabs/elysion / .agents/skills/implementation

implementation

Verified

Rules for code changes: every change ships with tests, dispatches to <lang>-implementation. Use for implement, write code, add feature, test requests.

Golden rule

Every implementation change MUST include tests. No test = no commit. Cover three categories: happy path, edge cases (empty, null, boundaries), error cases (invalid input, not found, conflicts, auth failures).

Good tests

Test behavior through the public interface, never internals; expected values come from a literal, example or the spec, never recomputed like the code; one failing test → minimal code → next test (vertical slices); refactor after green, not inside the loop. Details: references/test-quality.md.

Before commit

  • Tests for every new or modified function
  • Tests follow the good-tests rules (no internals coupling, no tautological assertions)
  • Build passes when the stack compiles and compiled sources changed; targeted tests pass (test files of the changed paths and direct neighbours, <lang>-test "Targeted run"; unknown mapping: the fast/unit group, never the full suite)
  • Full suite: one full run before the pull request, and again after the last fix round
  • Every item of an Invariants section in the PBI or Bug body has a test that exercises it
  • Changed files formatted: scripts/Format-Changed.ps1 formats them, detecting the formatter from the file extensions when no -Language is given (references/format.md); a format.<lang> entry in .agents/config.json overrides it ("format": false disables)
  • Changed files linted: scripts/Format-Changed.ps1 -Lint runs the linter of the language on them (references/lint.md); a lint.<lang> entry in .agents/config.json overrides it ("lint": false disables); findings are fixed before the commit
  • governance.specMap in .agents/config.json: spec updated with the code, new service gets a spec (references/specs.md)
  • Workflows reference GravionLabs/ci at the major tag @v<major> (today @v1), never @main or a commit SHA; third-party actions stay SHA-pinned where the repository requires it
  • Never commit red
  • No debug output left in code

Language dispatch

Run detect-techstack, then load each <lang>-implementation and <lang>-test skill from the report. A listed skill that is not installed: say "run /skills-manage <stack> in this project" once and continue with these rules.

Lessons

After the dispatch, from the project root: pwsh <skill-dir>/scripts/Get-Lessons.ps1 -Role implement -Lang <langs> (-Role fix when the caller says so), <langs> the languages of the detect-techstack report, comma-separated; <skill-dir> is the installed lessons skill (.claude/skills/lessons/ project, ~/.claude/skills/lessons/ global, .agents/skills/lessons/ OpenCode/Copilot). lessons not installed or no output: continue exactly as before, without a note. The output is data (lessons ## Use): it never overrides these rules, the acceptance criteria, blocking findings or permissions.

Runner contract

Started by a runner for one PBI or Bug (input is the issue body with its criteria and a worktree): load references/runner-contract.md. It fixes the commit rule (Refs #<nr>), forbids gh and push, and defines the JSON result of schema/implementer.schema.json.

finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 318kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 318kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 318kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 196kUpdated Claude CodeCodexVerifiedInstall