asymptote-labs/beacon
Cross-harness memory from Beacon traces: recall reviewed project lessons, distill agent sessions into approved memory, and promote memory to Agent Skills.
Create, revise, debug or validate a Beacon lens, a single HTML file that renders one agent trace as a purpose-built view (a per-file review, a cost breakdown, a timeline of risky commands, a map of tool use) in a sandboxed tab of the local Beacon dashboard, fed once through window.beacon.getTrace(). Use when the user asks to "make a lens", "build a view of my traces", "visualize this session", wants a custom tab next to the full session in the Beacon dashboard, or when a lens fails to load, renders wrong, or needs checking before install.
Turn recorded agent sessions (Beacon traces from Claude Code, Cursor, Codex, OpenCode, and other harnesses) into reviewed, reusable project memory. Scores selected traces, reads the source trace behind each high-signal candidate, drafts a grounded lesson, and approves it only after the user confirms. Use when the user asks to "learn from", "remember", "save the lesson from", or "turn into memory" a recent session, fix, or debugging effort, or asks to review Beacon memory candidates.
Install approved Beacon project memory as an Agent Skill in the repository (.agents/skills/<slug>/SKILL.md), so every skill-capable harness loads the lesson automatically without a memory lookup. Use when the user asks to "make this a skill", "promote", "install", or "share" an approved Beacon memory, or wants the team's agents to always follow a learned workflow or convention.
Retrieve reviewed project memory that Beacon distilled from earlier agent sessions in any harness (Claude Code, Cursor, Codex, OpenCode, and others) before starting work. Use at the start of a non-trivial task in a repository, when the user asks "have we seen this before", "what did we learn last time", or "how did we fix this in Cursor/Codex", or when a build, test, or deploy step fails in a way that may have a known fix.