proactive-context
Proactive context saving — handoff notes, Mess Hall logging, memory management
- License 许可
- MIT
- 最近更新
- 2026-11-12
- Stars
- ★ 81
- Forks
- 17
01 前言 Front Matter
- name
- proactive-context
- category
- core
- description
- Proactive context saving — handoff notes, Mess Hall logging, memory management
02 资源文件 3
- 许可证LICENSE1.0 KB
- 参考文档SOURCE.md369 B
- 参考文档TRUST.auto.yaml2.1 KB
03 SKILL.md 正文
Proactive Context Saving
Trigger: When context is getting tight, when brainstorming, when ending a session.
During Conversations
- Log ideas to Mess Hall (
11-Mess Hall/) during brainstorming — not just at the end - Cross-reference — note files, decisions, or threads that might be needed later
- Save as skill if a workflow repeats — don't clutter memory with procedures
Handoff Note Protocol
When ending a session or context is tight, write to 11-Mess Hall/YYYY-MM-DD-handoff.md:
# Session Handoff — [Date]
## Working On
- What we were doing when context ran out
## Decisions Made
- List of choices/decisions from the session
## Open Threads
- Things started but not finished
## Next Steps
- What to pick up on session resume
## Files Modified
- List of files that were changed
Agent Activation via Peer Transport
Context is no longer a passive accumulate-then-prune problem. hermes peer dm
spawns a real turn in the target agent — the active-memory layer. Wake another
agent only when there's something to act on:
hermes -p <my-profile> peer dm <target> "Pick up the open handoff at <path> — <task>"
- Handoffs (vault) = durable record; peer DM = live wake-up that proves the agent acted (synchronous reply).
- Each profile needs a distinct
api_server.port(gentech 8642, treasury 8643, gizmo 8644, pixel 8645) + a strongAPI_SERVER_KEY, then a gateway restart (from a separate shell — the gateway SIGTERMs children). - Bot Chat bloat pitfall: a peer DM targets the canonical "Bot Chat" session,
which doesn't auto-reset and can exceed the compressible context ceiling. Reset
it when you see "Context length exceeded":
hermes -p <profile> sessions rename <id> "__archived-bot-chat"thenhermes -p <profile> sessions archive --title "__archived-bot-chat" --yes. Recoverable, nothing deleted. - Full reference:
docs/peer-transport-and-activation.md.
Memory Bar Rule
Only show [? XX%] when memory is 80-100% full. Don't show it below 80%.
Session Start Protocol
- Read latest Mess Hall file for context
- Check
09-Green Room/ideas.mdand11-Mess Hall/considerations.md - Search sessions if user references something from before
- Don't ask the user to repeat themselves — find it in the vault first
Skill Over Memory
When discovering a repeatable workflow:
- Memory = facts (user preferences, environment details)
- Skill = behaviors (procedures, workflows, patterns)
- Save as skill if it'll be used again — skills persist better
Mess Hall as Scratchpad
The Mess Hall (11-Mess Hall/) is for:
- Ideas being fleshed out
- Cross-references to check later
- "Things to think about" notes
- Mid-conversation observations
- Session handoff notes
Don't wait until end of session. Log as you go.
04 同仓库更多 Skills
recipe-performance
Diagnose a specific MetaMask user flow's real-app performance and Sentry critical-user-flow telemetry with deterministic mm-harness recipes. Use for CUF boundaries, cold/warm/background-resume behavior, latency regressions, lifecycle tags, duplicate or blocking network requests, long tasks, layout/paint cost, screen redraws, React rerenders, or browser trace analysis where tests alone cannot prove the real app.
mexc-spot-websocket
MEXC Spot WebSocket workflows for public streams, protobuf handling, depth maintenance, listen keys, and private streams without exposing API keys.
mexc-spot-rest
MEXC Spot REST workflows for public data, signed account requests, orders, wallet/admin actions, and secure signing without exposing API keys.
mexc-futures-websocket
MEXC Futures WebSocket workflows for public streams, private account/order/position streams, ping/pong, subscriptions, and depth maintenance.
0xklen-version-and-refresh-sanctions-lists
暂无简介
verify-tests-cover-changed-lines
Use when a PR adds code that the existing suite may not exercise. Measures patch coverage — the share of newly changed lines hit by tests — before approving.