Skills for adopting Reflect — long-term memory for AI agents that gets smarter over time.
Walks a coding agent through adding Reflect long-term memory to a Python agent project end-to-end: SDK install, framework-specific loop placement, parameter tuning, LLM-as-judge wiring, and a mandatory smoke test that proves the loop closes.
Triggers on phrases like "add Reflect to my agent", "give my agent memory",
"wire up client.trace", "ctx.memories is empty", "set up an LLM judge
for Reflect".
Covers 5 Python frameworks: OpenAI Agents SDK, Claude Agent SDK, LangGraph, Pydantic AI, and a generic-loop fallback.
# Add a single skill globally
npx skills add StarlightSearch/reflect-skills@integrate-reflect -g -yBrowse all skills: https://skills.sh/StarlightSearch/reflect-skills
Reflect records what happened on each agent run (task, trajectory, outcome), distills it into a one-sentence reflection, and re-injects the most useful past lessons into future runs. Memories are ranked by a blend of semantic similarity and a learned utility score (q-value). Memories that lead to good outcomes float up; memories tied to failures sink until later success rehabilitates them.
📚 Docs: https://docs.starlight-search.com
🖥️ Console: https://reflect.starlight-search.com
⚡ Hosted API: https://api.starlight-search.com
For agents fetching docs: https://docs.starlight-search.com/llms-full.txt
(plain text, fetch with curl/urllib).
Open an issue or PR. Skills here are built TDD-style — each skill should be
backed by baseline scenarios that fail without it and pass with it. See
integrate-reflect/SKILL.md for an example of the structure.
MIT — see LICENSE.