domains / families / agent-tooling / nika

nika

combined

Runs repeatable AI work as checked, budgeted workflow files. The agent captures a repeated task as a .nika.yaml DAG, audits cost/permits/schema before a single token is spent, and runs it with tamper-evident traces. Runs repeatable AI work as checked, budgeted workflow files.

When to use

• The user asks to run, check, or author a `*.nika.yaml` workflow
• A task will be repeated (daily digest, triage, ETL, report, multi-step LLM
• The user wants a hard cost cap, a cost estimate before running, or
• A pipeline mixes models/providers (local + cloud) or mixes LLM steps with
• The user wants a run they can replay, verify, or reproduce later
• One-off questions or single tool calls — just answer or use a tool
• Autonomous code implementation/refactoring/PR review — use the `opencode` skill
• Interactive back-and-forth tasks — workflows are non-interactive by design

When not to use

• Static checks reduce risk but cannot prove that remote content, shell steps,
• Cost caps are not reliable for uncataloged paid models and a parallel wave
• Trace verification proves integrity of the recorded chain, not correctness

Skill page

Conflict

Both copies describe the same Nika workflow engine, but they are not interchangeable. The davepoon copy is a deep authoring guide focused on the 'native-first' law, the exec ledger, and specific builtin mappings (e.g., using nika:fetch instead of curl). The sickn33 copy is an operational guide focused on the Hermes agent context, installation, background execution, and specific error codes (NIKA-1704). The davepoon copy contains critical authoring constraints (the exec ledger) that the sickn33 copy lacks, while the sickn33 copy contains operational steps (background polling, installation) that the davepoon copy lacks. Combining them provides a complete skill for both authoring and running Nika workflows.