griddynamics/rosetta - Page 3

subagent-directives

MUST duties for a spawned subagent

Langue du texte source : anglais

tech-specs

To define clear, testable tech specs from requirements — target-state architecture, contracts, interfaces.

Langue du texte source : anglais

testing

To write thorough, isolated, idempotent tests — 80%+ coverage, external-only mocking, scenario-driven.

Langue du texte source : anglais

backlog

To validate story readiness for a sprint or to break work into a human WBS.

Langue du texte source : anglais

codemap

To generate, populate, and use a project codemap

Langue du texte source : anglais

coding-agents-hooks-authoring

To author, register, and test Rosetta hooks, add a SemanticKind, or debug a hook that won't fire.

Langue du texte source : anglais

coding

To implement features, fix bugs, and refactor with KISS/SOLID/DRY and systematic validation.

Langue du texte source : anglais

dangerous-actions

MUST activate when action OR its consequence is potentially dangerous/irreversible/destructive, or HIGH RISK, or environment is stage, pre-prod, prod, or blast radius.

Langue du texte source : anglais

data-collection

To gather QA source artifacts from issue tracker, wiki, test management system

Langue du texte source : anglais

debugging

To investigate errors, test failures, and unexpected behavior — root cause before fix.

Langue du texte source : anglais

design

To decide architecture — alternatives, tradeoffs, committed decision with rationale, boundaries, domain-specific concerns.

Langue du texte source : anglais

deviation

MUST activate when intent is unclear, or cannot follow or deviate from original intent, or easily or reliably solve the problem, something came as SURPRISE or UNEXPECTED, you cannot bet $100 on solution, unknowns or assumptions critically affect solution, you…

Langue du texte source : anglais

discovery

To establish what exists — affected areas, behavior, prior attempts, dependencies, data.

Langue du texte source : anglais

hitl

MUST activate for ALL tasks — planning, execution, validation, review: session-wide work with human together, human-in-the-loop questioning, approvals, stop-and-wait vs proceed. NEVER assume approval. MANDATORY unless user requested EXACTLY `fully autonomous`…

Langue du texte source : anglais

large-workspace-handling

To partition large workspaces (100+ files) into scoped subagent tasks when context is insufficient.

Langue du texte source : anglais

load-project-context

To load the project's business, behavioral, and technical context.

Langue du texte source : anglais

natural-writing

To rewrite text in a clear, natural, honest human tone — no AI slop, hype, or robotic phrasing.

Langue du texte source : anglais

planning

To build execution-ready AI sessions graph with intent/ACs/checklists.

Langue du texte source : anglais

post-mortem

Diagnose instruction defects and optionally submit Rosetta GitHub issue

Langue du texte source : anglais

qa-knowledge

To run QA engineering — requirements/gap analysis, scenario & spec design, test implementation, failure triage — over the QA knowledge base.

Langue du texte source : anglais

qa-structure

To resolve QA session paths, identifiers/slugs, and state-file shape

Langue du texte source : anglais

questioning

To ask targeted clarification questions only when high-impact unknowns block safe execution.

Langue du texte source : anglais

reasoning

To apply structured meta-cognitive thinking to complex problems. Invoke on explicit only.

Langue du texte source : anglais

requirements-authoring

To author, update, and validate functional/non-functional requirements as atomic units with user approval.

Langue du texte source : anglais

requirements-use

To consume approved requirements for planning, implementation, and validation, with traceability and HITL.

Langue du texte source : anglais

research

To run systematic deep research via meta-prompting — grounded references, incremental tracking, self-validation.

Langue du texte source : anglais

reverse-engineering

To reverse-engineer code into a spec: extract behavior and domain logic — WHAT and WHY, not HOW.

Langue du texte source : anglais

risk-assessment

MUST activate before execution when environment has access to databases, cloud services, S3, or similar external systems, any access to servers of higher environments, and when assessing environment risk level.

Langue du texte source : anglais

rosetta

Rosetta identifies and routes user request to the most matching workflow

Langue du texte source : anglais

security

Run authorized, evidence-preserving security reviews and prepare remediation inputs.

Langue du texte source : anglais

self-learning

MUST STOP and activate on: any failure/error, mistakes, wrong/unexpected result, expected≠actual mismatch, 2 consecutive mismatches, unhappy/upset user, user WRITING IN CAPS, something didn't work. No edits, no reverses, no actions, except this skill.

Langue du texte source : anglais

self-organization

MUST activate at 65%+ context usage or on large-file restructuring. Proactive planning, reorganization, stale-content cleanup.

Langue du texte source : anglais

sensitive-data

MUST activate when you suspect, there is a slight chance, encounter, read, process, or are about to output any sensitive or possibly sensitive data including PII, PCI, HIPAA, PHI, GDPR, SOC2, FedRAMP, secrets, API keys, passwords, credentials, tokens,…

Langue du texte source : anglais

solr-extending

To build Solr plugins: SearchComponent, QParser, URP, DocTransformer.

Langue du texte source : anglais

solr-query

To build and debug Solr queries: eDisMax, block join, JSON facets, kNN, explain.

Langue du texte source : anglais

solr-schema

To design and audit Solr schemas: field types, analyzers, docValues, solrconfig.

Langue du texte source : anglais

solr-semantic-search

To build Solr phrase-tagging semantic search: concept tagging, taxonomy, graph paths.

Langue du texte source : anglais

specflow-use

To connect Rosetta with Grid Dynamics SpecFlow MCP; only when SpecFlow is mentioned and the MCP is installed.

Langue du texte source : anglais

subagent-directives

MUST duties for a spawned subagent

Langue du texte source : anglais

tech-specs

To define clear, testable tech specs from requirements — target-state architecture, contracts, interfaces.

Langue du texte source : anglais