Skip to content

feat: add algobot-cli skill for Agent Studio management#7

Open
PLNech wants to merge 3 commits intomainfrom
feat/algobot-cli-skill
Open

feat: add algobot-cli skill for Agent Studio management#7
PLNech wants to merge 3 commits intomainfrom
feat/algobot-cli-skill

Conversation

@PLNech
Copy link
Member

@PLNech PLNech commented Mar 19, 2026

Summary

New skill for managing Algolia Agent Studio agents from the terminal via algobot (npm: algobot-ai).

  • Triggers on: create/list/update/publish/delete agents, chat with agents, config-as-code deployments, multi-env management
  • Does NOT trigger on: Algolia search index ops (algolia-cli), general LLM tasks

What's included

  • SKILL.md (152 lines) — setup, non-interactive mode gotchas, core commands, config-as-code workflow, global flags
  • references/commands.md — full command reference for every subcommand
  • references/config-as-code.md — templates, variables, multi-env, CI/CD patterns

Key gotchas captured

  • TUI requires TTY — agents must use ask or --text mode
  • --dry-run is API-layer enforced, not just a flag
  • --jq is built-in, no need to pipe to jq
  • Exit codes are 0/1 only in v2.0

PLNech added 3 commits March 19, 2026 17:14
- Broaden description to trigger on RAG, conversational experiences,
  genAI content generation (carousels, headers, product descriptions)
  even when user doesn't mention algobot/Agent Studio by name
- Add Agent Studio Capabilities section: memory tools (algolia_memorize,
  algolia_ponder, algolia_memory_search), MCP tool integrations, and
  experimental features (citation markers, date injection, caching)
- Add gotcha: don't fabricate REST endpoint URLs — use algobot agents get
- Add explicit algobot tools list/add/remove for MCP configuration
- Add 9 evals covering RAG, conversational, genAI carousel, multi-env
  deploy, memory, GDPR, MCP tools, InstantSearch coexistence, citations
- Validated via two eval iterations (91.9% pass rate in iteration 2)
Switch from passive "Use this skill whenever..." to assertive ALWAYS +
explicit trigger keyword list. Addresses undertriggering in skill
selection — makes trigger intent unambiguous even when user doesn't
name Agent Studio or algobot explicitly.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant